217: Anthropic Just Dropped Their Internal Data Playbook (copy this)

Speaker: 00:00:00

Avery Smith-3: So Anthropic, the makers

of Claude, literally just dropped an

2

: 00:00:03

absolute masterclass on how they analyze

data internally, and they posted a blog

3

: 00:00:08

post that is four thousand five hundred

words, and there's a lot in there.

4

: 00:00:12

So I summarized that entire blog

post, and I will explain it to you

5

: 00:00:16

like you're five years old in today's

episode, and literally you can steal

6

: 00:00:19

it and learn how to analyze data

just like a Claude data analyst.

7

: 00:00:23

So this is what Claude

is actually claiming.

8

: 00:00:26

They're claiming that they now

do self-serve analytics, which

9

: 00:00:29

is kind of a funny phrase.

10

: 00:00:30

Basically, it means allowing non-technical

people, non-data analysts to do data

11

: 00:00:36

analytics in easy ways, and this has

been a thing for the last decade or so.

12

: 00:00:41

In fact, it's one of the main reasons

why Tableau and Power BI became

13

: 00:00:44

so important with dashboards is it

allows business people, non-technical

14

: 00:00:48

people to actually kind of analyze

their data in predefined ways.

15

: 00:00:52

It's been really hard to do for

the last ten, fifteen years.

16

: 00:00:55

Now, basically, Anthropic just

tweeted that they are able to do

17

: 00:01:00

ninety-five percent accuracy on

all of their business analytics

18

: 00:01:05

queries with Claude, which is crazy.

19

: 00:01:08

That basically means that w- if anyone has

some sort of an analytics question, they

20

: 00:01:13

can answer it now with ninety-five percent

accuracy using this internal playbook.

21

: 00:01:17

So what are they actually doing, and

how can you replicate it in your own

22

: 00:01:21

organization, or how can you bring

this to an interview to make you a

23

: 00:01:25

more marketable aspiring data analyst?

24

: 00:01:28

So- Basically, like I said, self-serve

analytics has always kind of sucked.

25

: 00:01:32

It's when non-technical people are

analyzing the data sets, and there's

26

: 00:01:35

basically two different ways to do it.

27

: 00:01:37

Option A is you open up to everyone,

which basically means you have

28

: 00:01:40

non-data analyst people trying to

analyze data, and a lot can go wrong.

29

: 00:01:44

You can get really messy,

different queries, maybe

30

: 00:01:47

messy dashboards, conflicting

definitions, those type of things.

31

: 00:01:50

Or you lock it all down, which basically

means that, uh, you create a bajillion

32

: 00:01:55

different types of dashboards, but it

never really answers anyone's question

33

: 00:01:58

when they want it the way they want it.

34

: 00:02:00

And, uh, that's been, that's

been tricky in the past.

35

: 00:02:02

So now there's AI, and now you

can give, you know, Claude…

36

: 00:02:07

You can give someone Claude or

ChatGPT and point it to a database,

37

: 00:02:10

and you can have them ask ChatGPT

or Claude questions to the database.

38

: 00:02:15

Uh, but there's a big issue.

39

: 00:02:16

Number one, that we all think the

AI doesn't hallucinate, doesn't

40

: 00:02:19

lie, doesn't make things up, and

it does, and it can be wrong.

41

: 00:02:22

Uh, and number two, it gives everyone

like, "Oh, this is a hundred percent

42

: 00:02:26

accuracy," but it's, it's not, and

that can cause a lot of issues.

43

: 00:02:30

So, um, you know, AI is a great

solution for self-serve analytics, but

44

: 00:02:33

it causes a lot of problems as well.

45

: 00:02:36

So how did, uh, Anthropic

actually solve it?

46

: 00:02:38

Because what they're claiming

that ninety-five percent of their

47

: 00:02:40

business analytics queries are now

automatedly solved by Claude, and

48

: 00:02:44

they're ninety-five percent accuracy,

accurate, um, which is a big claim.

49

: 00:02:48

Like that's basically like,

"Hey, Claude is now our company

50

: 00:02:52

data analyst, essentially."

51

: 00:02:54

Now, I, I will mention here, um, that

the data team can now work on bigger

52

: 00:02:59

and better problems that are like

less sequel monkey questions, right?

53

: 00:03:03

Um, so it's not like they're

getting rid of their data

54

: 00:03:05

analyst or their data scientists.

55

: 00:03:07

It's just you don't have to do as

many ad hoc reportings, and you can

56

: 00:03:10

just focus on more important things.

57

: 00:03:12

And just managing this Claude

infrastructure of creating this

58

: 00:03:16

company-wide, uh, self-serve

analytics platform is a beast, and

59

: 00:03:20

we'll get to that here in a second.

60

: 00:03:22

Um, basically, in this article,

their thesis is data is very

61

: 00:03:25

different than software.

62

: 00:03:26

If you've, you know, heard about Claude

or Codex, um, for programming and software

63

: 00:03:30

engineering, it can do those things

really, really well out of the box.

64

: 00:03:34

Um, because coding has

lots of right answers.

65

: 00:03:37

There's ways to test things.

66

: 00:03:38

There's documentation that goes with code.

67

: 00:03:40

Um, and all those, you know,

infrastructure can basically

68

: 00:03:44

catch hallucinations.

69

: 00:03:45

It's a more solved problem.

70

: 00:03:47

Analytics, it's quite a bit

different because there's only one

71

: 00:03:50

right answer, and you don't really

know what the right answer is.

72

: 00:03:52

There's no way to actually test

what the answer is versus i-in

73

: 00:03:56

programming, you're like, "Does this

box open up if I click the button?"

74

: 00:03:59

You can test that.

75

: 00:04:00

There's no way to know, like if I ask

Claude for the m- you know, the mean

76

: 00:04:03

of our sales over the last month, you

really have to like go actually run the

77

: 00:04:07

query yourself To make sure that Claude's

not giving you, uh, a false answer.

78

: 00:04:12

So, um, their, their argument

is we're not having issues

79

: 00:04:15

coming up with code generation.

80

: 00:04:17

It's basically all of the context

and verification that goes around

81

: 00:04:21

solving a business analytics problem.

82

: 00:04:23

And LLMs historically have been pretty bad

at this, uh, for a multitude of reasons.

83

: 00:04:27

One is that we give it unclear directions.

84

: 00:04:30

I don't know about you guys, but if

you're anything, uh, like me, you don't

85

: 00:04:33

necessarily give Claude or ChatGPT the

most specific instructions on planet

86

: 00:04:37

Earth, and there's some ambiguity.

87

: 00:04:39

And the problem with that is, like,

it can go into the database and, like,

88

: 00:04:43

it thinks it knows what you're talking

about, but it finds a different column,

89

: 00:04:46

or it's not using the same definition.

90

: 00:04:48

You're not basically on the

same page as ChatGPT unless you

91

: 00:04:51

give really explicit directions.

92

: 00:04:53

Number two, there's data staleness,

which basically means that your database

93

: 00:04:57

is constantly changing, uh, over time.

94

: 00:04:59

Definitions change, tables change,

and, uh, these AI LLMs, they're not

95

: 00:05:04

really good at following with that.

96

: 00:05:05

Like, they don't have the business

context, the domain context that you

97

: 00:05:09

may have as a human being on the other

side of like, "This is why we made

98

: 00:05:12

those changes," you know, "This is

why it's better," so on and so forth.

99

: 00:05:15

And then number three is it just doesn't

know where to find the right thing.

100

: 00:05:19

Like, it thinks the data's in there,

it's looking, but it's not entirely sure

101

: 00:05:22

Avery Smith-4: So here's what

Anthropic did to try to solve this

102

: 00:05:24

problem, and they're calling it

Anthropic's Agent Analytics Stack.

103

: 00:05:28

And there's basically four

different stages right here, and

104

: 00:05:31

each one is built to try to take

one of those previous problems

105

: 00:05:35

that we talked about and solve it.

106

: 00:05:36

So the first one is data foundations,

and basically, it just means you

107

: 00:05:39

have really solid data foundations.

108

: 00:05:42

It means you're very clear on what

a table is, what it actually has,

109

: 00:05:45

what a row represents, what a column

represents, and how often it's updated.

110

: 00:05:49

Um, number two is you only have one

source of truth, and the idea is

111

: 00:05:53

if you have a sales table in your

database, you don't have, like,

112

: 00:05:56

another sales table in your database.

113

: 00:05:58

Like, there's only one sales

table, and that is the sales table.

114

: 00:06:02

There are no other sales tables.

115

: 00:06:04

And for some of you guys listening

who might be more junior data analysts

116

: 00:06:07

or aspiring data analysts might be

thinking, "Well, that makes sense.

117

: 00:06:09

Why would it ever be a different case?"

118

: 00:06:11

And the issue is when you get to, like,

large organizations, something like

119

: 00:06:14

Anthropic or when I worked at ExxonMobil,

you gotta think that there's literally

120

: 00:06:18

seventy thousand plus employees, and all

of them might need access to that table,

121

: 00:06:23

and they might need it slightly different.

122

: 00:06:25

So you might have someone that's

like, "Oh, this is their sales

123

: 00:06:28

table, but we only need the weekly

averages," so they create, you know,

124

: 00:06:31

the weekly average sales table.

125

: 00:06:34

And then there's someone else who's like,

"Oh, well, we actually only need the

126

: 00:06:36

sales from Monday, Wednesday and Friday,"

and so they create this other table.

127

: 00:06:39

And basically, you just get a bajillion

versions of really the same table.

128

: 00:06:43

So, uh, one source of truth,

really important here.

129

: 00:06:46

Number three, they develop skills.

130

: 00:06:48

These are like Claude skills for

LLMs that specifically do a repeated

131

: 00:06:52

task with specific instructions and

maybe even some, uh, accompanying

132

: 00:06:57

code to make it really repetitive.

133

: 00:06:59

LLMs have inherent

randomness built into them.

134

: 00:07:03

They are non-deterministic, as in

you don't get the answer every time,

135

: 00:07:07

the same answer every time you ask

the same question, and skills helps

136

: 00:07:10

make it more deterministic, that

there actually is a specific answer.

137

: 00:07:14

This is exactly what you should be doing.

138

: 00:07:16

So it's basically like instructions and

almost code files to actually follow

139

: 00:07:19

every single time this gets asked.

140

: 00:07:21

And the fourth one is validation, and

that is making sure that the LLMs are

141

: 00:07:25

actually doing what you think they

are and validating their answers.

142

: 00:07:28

So let's dive in a little bit deeper.

143

: 00:07:30

So like I said, uh, layers one and layers

two, basically this is just having good

144

: 00:07:35

data governance and good data foundations.

145

: 00:07:37

One source of truth.

146

: 00:07:38

Um, they also make sure that they have

like little, uh, descriptions for each one

147

: 00:07:42

of your different tables that describes

what the table is and what it isn't.

148

: 00:07:46

Uh, you know, LLMs are really good

at reading text, so if you add a

149

: 00:07:49

little bit of text with your tables

that explains what's going on, the

150

: 00:07:52

LLM understands the context a little

bit better versus just looking at the

151

: 00:07:55

rows and the columns and guessing.

152

: 00:07:57

Um, you can think of this as

like a README file for your data.

153

: 00:08:01

In code, in building software, in

software engineering, in programming,

154

: 00:08:05

we've always had README files.

155

: 00:08:06

If you're unfamiliar, a README

file, you can just think of it

156

: 00:08:08

as like a summary of the actual

what's going on in your code base.

157

: 00:08:13

Like all of these different folders,

all these different files, all these

158

: 00:08:15

different code scripts, what's going on.

159

: 00:08:17

So it's just a human way to

describe what's going on for

160

: 00:08:20

your code or your different, you

know, databases in this case.

161

: 00:08:24

And they also feed it

company knowledge maps.

162

: 00:08:26

So for this system, they give it roadmaps,

org charts, decisions, so like a bunch

163

: 00:08:31

of business context that isn't data.

164

: 00:08:33

It's not data related.

165

: 00:08:34

It's all business and domain related,

but that extra information helps the

166

: 00:08:39

LLMs make smarter choices on how to

actually analyzing the da-- how to

167

: 00:08:43

analyze the data based off of what

the, what the context says So they

168

: 00:08:47

actually tried an experiment here,

which I thought was really interesting,

169

: 00:08:50

where they basically took all the data

analysts' and all the data scientists'

170

: 00:08:54

old sequel files, and they said, "Here,

Claude, you know, learn from these.

171

: 00:08:59

These are all, all the things that

our engineers and our analysts and our

172

: 00:09:02

data scientists have done over time.

173

: 00:09:03

Uh, learn from it."

174

: 00:09:05

And it actually didn't really

help, which was really interesting.

175

: 00:09:08

Um, it didn't know what code to use when.

176

: 00:09:12

Um, and they found that there

was a right answer eighty percent

177

: 00:09:14

of the time, but Claude wasn't

good at pulling that answer out.

178

: 00:09:17

And so what's actually been the biggest

skill, uh, uh, I guess the biggest,

179

: 00:09:22

uh, unlock is actually having skills.

180

: 00:09:24

And that went from twenty-one

percent accuracy in actually

181

: 00:09:27

analyzing data to ninety-five

percent accuracy in analyzing data.

182

: 00:09:31

And if you're unfamiliar with,

like, what a Claude skill is, or

183

: 00:09:33

I think they have some equivalent

in ChatGPT and OpenAI and Codex.

184

: 00:09:37

But basically a, an LLM skill,

an AI skill is a reusable

185

: 00:09:41

step-by-step pattern to follow.

186

: 00:09:43

Think of it almost like a recipe for

AI LLM models to actually follow.

187

: 00:09:48

So like I said, majority of the

time they're written kind of like

188

: 00:09:51

a human would write them, and it's

just like, "Hey, AI, do exactly this.

189

: 00:09:55

Step one, step two, step three.

190

: 00:09:57

Look out for this.

191

: 00:09:58

Be aware of this."

192

: 00:09:59

And it might have some coding files

specifically like, "This is what your code

193

: 00:10:02

should look like if you generate code."

194

: 00:10:04

Um, so they…

195

: 00:10:06

It, it's, it's essentially what a

senior analyst's thoughts written down

196

: 00:10:10

on paper, uh, for a specific task.

197

: 00:10:12

So you might have a skill on how

to, you know, create a, a bar, a

198

: 00:10:16

bar chart, or you might have a skill

on how to do a hypothesis test or

199

: 00:10:20

AB testing or something like that.

200

: 00:10:22

And it's basically like you have

your, your team get together and write

201

: 00:10:25

down exactly what the process is.

202

: 00:10:26

It's like a standard operating procedure

that you'd give to a junior analyst,

203

: 00:10:29

"Hey, follow this," except for now the

junior data analyst is Claude or an AI One

204

: 00:10:34

issue they saw was if you don't actually

update these skills, like if you don't

205

: 00:10:38

like constantly add to them and improve

them, that the accuracy slides over time.

206

: 00:10:43

They actually were at ninety-five

percent accuracy, and then they

207

: 00:10:45

jumped down to sixty-five percent

accuracy in only a few weeks.

208

: 00:10:49

Um, so you need to make sure

you're updating your skills.

209

: 00:10:51

And the last thing is they wanted to make

sure that their skills were everywhere.

210

: 00:10:54

So analytics is really changing.

211

: 00:10:56

Uh, and this-- You probably haven't

seen this in big organizations now.

212

: 00:11:00

It's just kind of rolling out to

maybe, you know, these more frontier

213

: 00:11:03

trillion-dollar companies, um, and maybe

like small solopreneurs like, like me.

214

: 00:11:08

Um, but the way that we do

data analytics is changing.

215

: 00:11:10

So obviously, like in the past, you'd use

Excel to do data analytics, and there's

216

: 00:11:14

still literally billions and billions of

Excel files that we will analyze in Excel.

217

: 00:11:20

Uh, but gradually, you know, ten,

fifteen years down the road, I'm

218

: 00:11:23

not sure if that will be the case.

219

: 00:11:25

We will probably be analyzing data

in a different way than we are now.

220

: 00:11:28

And before you're really scared and like,

"Oh my gosh, this is awful, AI's coming

221

: 00:11:31

for my job," well, just think about this.

222

: 00:11:33

Uh, basically, Power BI

came out fifteen years ago.

223

: 00:11:36

So fifteen years ago, there were

like basically no dashboards.

224

: 00:11:40

Tableau was around, but not super popular

at the time, yet it was about to be.

225

: 00:11:43

Uh, about twenty eighteen it

started to get really popular.

226

: 00:11:46

So it's just like, yes, the way that

we analyze data changes over a decade.

227

: 00:11:50

That's the truth.

228

: 00:11:51

Um, and just know that right now we are

moving into, you know, analyzing our data

229

: 00:11:55

with these chatbots, and those chatbots

may be in multiple different places.

230

: 00:11:59

So for example, at my company, um, I try

to analyze data on, you know, my YouTube

231

: 00:12:04

watches or my podcast listens, and I've

been trying to tr- to automate that as

232

: 00:12:09

much as I can or make it easier for me

to follow, you know, all these analytics.

233

: 00:12:13

And so we actually have a bot that

will help me with these analytics where

234

: 00:12:16

I can just ask it natural language

questions like, "How many, uh, views

235

: 00:12:20

did the last YouTube video get?"

236

: 00:12:21

You know, "How many listens

did this podcast episode get?"

237

: 00:12:25

And we can actually do that on a website

that I've built and also in our Slack.

238

: 00:12:28

So they want to make sure that they

have the truth and those-- these skills

239

: 00:12:32

avail-available everywhere, whether

it's, you know, you're coding, whether

240

: 00:12:35

you're using like a website or a

dashboard or whether you're in Slack.

241

: 00:12:38

So those are the keys to having

good skills in your organization.

242

: 00:12:42

And the last thing is, even

if it has a good skill, how

243

: 00:12:45

do you know that it's correct?

244

: 00:12:46

And that's what we call verifications.

245

: 00:12:48

And so what, what Anthropic's doing, what

Claude's doing is for any analytics they

246

: 00:12:52

do, they have the sources in the footer.

247

: 00:12:54

Like this is where we

got this information.

248

: 00:12:57

This is how we calculate it.

249

: 00:12:58

This is the table we used.

250

: 00:13:00

Um, so that way it's like very clear

that you could look at the table and

251

: 00:13:03

be like, "Oh, that is the right table,"

or, "It's not even the right table."

252

: 00:13:06

They also have a freshness

and a version stamp on every

253

: 00:13:09

data model and how old it is.

254

: 00:13:11

So like think about like i- if

your data changes over time.

255

: 00:13:15

They're basically timestamping

everything, so that way you know,

256

: 00:13:18

okay, we can trace it back to this

database on this day type of a thing.

257

: 00:13:22

Uh, they're also doing correction

harvesting, which is a really fancy way

258

: 00:13:25

to say they're giving the AI feedback.

259

: 00:13:27

So every time that this Claude

data analyst gets something wrong,

260

: 00:13:32

the humans are saying, "Hey,

you actually did this wrong.

261

: 00:13:35

You know, you're supposed to

grab from database A, and you

262

: 00:13:37

grabbed it from database B."

263

: 00:13:38

Or maybe you, you know, you

did your query wrong some way.

264

: 00:13:42

And every time that feedback goes

from the human to the agent, the agent

265

: 00:13:47

actually updates itself, and it's

like, "Oh, okay, I'm gonna mark that

266

: 00:13:51

as something to try in the future."

267

: 00:13:53

And the last thing they add is basically

before it gives any answer back to the

268

: 00:13:58

human, they run a second agent against

it that's called an adversarial review.

269

: 00:14:02

And basically, if, if you are the AI

data analyst and you come up with an

270

: 00:14:06

answer and you're like, "The average over

the last, you know, the average revenue

271

: 00:14:09

over the last month was thirty thousand

dollars," this ad-adversarial re-review

272

: 00:14:13

comes in and says, "Is it though?

273

: 00:14:15

Like, does, does that actually make sense?

274

: 00:14:17

Uh, like, it's been this for the last

month and this for the last month.

275

: 00:14:20

Are you a hundred percent sure?"

276

: 00:14:22

Um, it's basically trying to prove

the first agent incorrect before

277

: 00:14:27

actually giving them the model,

the information to the human.

278

: 00:14:30

So that way, it's like almost like a peer

review, a double check from an agent to

279

: 00:14:33

actually make sure that the analytics

is correct So this might be really

280

: 00:14:38

interesting to some of you guys, and this

might be really scary to some of you guys.

281

: 00:14:41

You're like, "Oh my gosh, these

AI agents are coming for my job."

282

: 00:14:44

Well, the first thing I'll tell you

that none of this is actually new.

283

: 00:14:47

It's just kind of packaged

in a fancy prettified way.

284

: 00:14:50

Like, if you literally take AI out of

this, it's just pure data fundamentals,

285

: 00:14:55

things that we've had for decades.

286

: 00:14:57

We've talked about this for years.

287

: 00:14:58

Like, yes, it's good to

have good data quality.

288

: 00:15:01

Yes, it's good to have good data

governance, like to actually know

289

: 00:15:04

what, what tables mean and what

columns mean and what rows mean.

290

: 00:15:08

Yes, we should repeat

our analysis when we can.

291

: 00:15:10

If we can analyze the data in a

uniform way, we should do that.

292

: 00:15:14

And yes, we should have verification.

293

: 00:15:16

Like, if I do an analysis,

someone else should check it

294

: 00:15:18

to make sure it all looks good.

295

: 00:15:20

This is not new.

296

: 00:15:21

It's just AI-fied, essentially.

297

: 00:15:24

The next step is this is actually a ton

of work to do, and really I don't see, you

298

: 00:15:29

know, a whole lot of companies being able

to pull this off bec- other than, like,

299

: 00:15:33

Anthropic, for example, because Anthropic

has literally trillions of dollars.

300

: 00:15:38

Uh, you know, they're growing like crazy.

301

: 00:15:39

They have tons of employees.

302

: 00:15:40

But all that documentation, all that

governance, all that quality, all

303

: 00:15:44

that metric mapping and, you know,

adding all the business information

304

: 00:15:47

to Claude, it takes hundreds of hours.

305

: 00:15:50

It takes so much time.

306

: 00:15:51

Before we even talk about maintenance,

like we talked about how they slipped

307

: 00:15:54

from ninety-five percent accuracy

to sixty-five percent accuracy

308

: 00:15:56

by not maintaining their skills.

309

: 00:15:58

Like, there's so much upfront

work and so much maintenance

310

: 00:16:02

work on this that it's insane.

311

: 00:16:04

I'm not the only person

who actually noticed this.

312

: 00:16:06

Uh, Kristen Lum said, "This work takes

hundreds and hundreds of upfront hours

313

: 00:16:10

at any moderately sized organization, and

that's not even counting maintenance."

314

: 00:16:14

So there is tons of work to be done

even if this is working, even this is

315

: 00:16:19

set up, you know, at normal companies.

316

: 00:16:21

I mean, I'm not Ex-ExxonMobil.

317

: 00:16:23

I haven't been at

ExxonMobil in five years.

318

: 00:16:25

I have no clue where they're at.

319

: 00:16:26

I have no insight.

320

: 00:16:27

A lot of people that I knew

there no longer work there.

321

: 00:16:29

But, like, just like the security

and privacy- concerns that

322

: 00:16:35

Exxon would have about all of

this would take years to solve.

323

: 00:16:37

Not, not even like

implementing and setting it up.

324

: 00:16:40

Maybe that's changed, I don't know.

325

: 00:16:41

But my point is these large

organizations, even ones with

326

: 00:16:44

billions of dollars, this is gonna

be difficult for them to pull off.

327

: 00:16:47

Um, the crazy thing about all this

is they literally just gave this out.

328

: 00:16:52

It's like they literally give you a skill

sheet, um, a skill file that you can

329

: 00:16:56

literally just copy and use for your own

personal analysis, or you can use it on

330

: 00:17:00

your team and organization's analysis.

331

: 00:17:02

Um, I have a little part of it right

here, or you can just go to the

332

: 00:17:04

blog post and find the full file.

333

: 00:17:08

My point here, though, is with all these

jobs are- with all these things that we

334

: 00:17:12

have to be doing for AI to become a good

data analyst, it's like Anthropic's not

335

: 00:17:17

getting rid of the data analyst right now.

336

: 00:17:18

They have four hundred roles open, and

eight of them at least are in data.

337

: 00:17:23

They have four thousand seven hundred

and forty-two employees on, on

338

: 00:17:25

LinkedIn and, uh, one- one thousand

four hundred and seventy-eight of

339

: 00:17:29

them deal with data, and a hundred and

ninety-six of them are data analysts.

340

: 00:17:33

So if this company that has mastered

ninety-five percent accuracy, the AI data

341

: 00:17:37

analyst is still hiring data people, I

think that AI jobs aren't going away.

342

: 00:17:41

Like, this is the company

that if they could get rid

343

: 00:17:44

of humans, they would, right?

344

: 00:17:45

If you've heard the CEO talk about

it, he thinks it's happening,

345

: 00:17:48

and you don't really see that

in their hiring numbers yet.

346

: 00:17:52

Um, my point of view is like this is

literally going to free you up to do

347

: 00:17:55

higher value work, including creating

and maintaining systems like this.

348

: 00:17:59

Like, like I said, like you guys

as data analysts are the people

349

: 00:18:04

best suited for the AI period.

350

: 00:18:06

Like, you guys know numbers, and

if you can compare numbers with

351

: 00:18:10

AI, you're going to be undefeated.

352

: 00:18:12

You're gonna be employed for a really

long time, and just the fact that you're

353

: 00:18:15

listening to this right now tells me

you're one of those people because

354

: 00:18:18

you're interested in data, you're

interested in AI, and if you can really

355

: 00:18:20

carve a niche that's AI plus data, I

think you're gonna land an awesome job.

356

: 00:18:25

I think you're gonna get

promoted to an awesome job.

357

: 00:18:27

I think you're gonna make a lot of money

in your career for a really long time.

358

: 00:18:31

So if you found this fascinating,

my name's Avery Smith.

359

: 00:18:34

Please hit subscribe because I really

want to talk about how data and AI

360

: 00:18:37

intertwine over the next six months,

and I want you to be on this journey.

361

: 00:18:41

I will see you in the next episode.