Robert Otani of CORE Studio at Thornton Tomasetti joins the show to talk about about his journey from a structural engineer in the 90s to leading the CORE Studio at Thornton Tomasetti. This episode covers the importance of computational design, BIM, and the impact of AI and machine learning in the engineering and architecture industries. Rob also shares insights from their recent AEC hackathon, the creation of CORE Studio, and innovative projects like developing an AI tool by capturing the depth of knowledge shared in the emails of a late colleague.

Episode Links:

• Connect with Rob Otani on LinkedIn
• Thornton Tomasetti’s CORE group

Watch this episode on YouTube or Spotify.

-----

The Confluence podcast is a collaboration between TRXL and AVAIL, and is produced by TRXL Media.

• Learn more about them at https://trxl.co/media and https://getavail.com
• Learn more about the Confluence podcast at https://confluence.getavail.com/podcast
• Learn about the Confluence events at https://confluence.getavail.com

Welcome to another Confluence podcast.

2

I'm Randall Stevens and I've

got Evan Troxel here with me.

3

And our guest today is Rob

Otani from Thornton Tomasetti.

4

Uh, just to, I'll give a brief intro,

Rob, and then I'll let you, you know,

5

fill everybody in in more detail.

6

But, uh, Rob is the CTO and is a

managing partner at Thornton Tomasetti.

7

And he runs what's called the

CORE Studio, which I'm sure

8

we'll talk quite a bit about.

9

But he's got a great group

of development team there.

10

Thanks Uh, at Thornton Thomas

City that do a lot of great work.

11

I think you all just recently

held your infamous hackathon.

12

So maybe we can talk a little bit

about that and how all that went, but,

13

uh, welcome to the podcast.

14

Oh, and I didn't say that Rob

participated in our, in our fall

15

confluence event here in Lexington.

16

So it was great to have you here and the

group really enjoyed the conversation.

17

So we're going to try to get

this out to a broader audience.

18

So welcome to the podcast.

19

Robert Otani: Thank you.

20

Randall Stevens: So, yeah, maybe just

a little bit more of your background

21

and then we can kind of dive into

what your, your team's mission

22

is there at Thornton Tomasetti.

23

Robert Otani: Yeah, so, you know,

I'm CTO now, but, uh, I started out

24

as a, you know, innocent structural

engineer back in the nineties.

25

Um, and, um, you know, um, you know,

felt like I was, uh, you know, pretty

26

high performing production engineer.

27

And, um, in about 2007, I believe.

28

I was, uh, involved in an organization

called SCIONI, which is the Structural

29

Engineers Association of New York,

which I actually, I think at that

30

time I was either president or,

or close president elect maybe.

31

Um, and I kept hearing about

computational design around that

32

time, and people were using something

called generative components.

33

And, which I never even opened or tried.

34

Um, So I decided I was part of the

programs committee and decided to,

35

um, organize with other people, uh,

that were part of the, uh, uh, team,

36

an event on computational geometry.

37

That was the name of the event.

38

And we invited some people that are

now like, you know, pillars of the

39

computational design industry, which

was Ian Keogh, was that Gerhapold?

40

Um, Dave Fano and, uh,

the, who else was there?

41

He was at Shop Architects at the time.

42

Dave, um, Josh Amig, I think Steve

Sanderson, they're all sort of the case

43

design team that ended up going to WeWork

and started other, other companies aside

44

from that, um, Oner Goon, um, who's now

at, I think he's at Reebok, I believe.

45

Um, but was at KPF at the time, and then,

you know, one of the, I would say the,

46

the, the, the fathers of competition

design was Neil Katz at SOM and, um,

47

Kat Park as well, who's, who's, who's

in the San Francisco office, who I

48

just saw, she came to our conference.

49

Um, so that when I saw that conference,

you know, and the things that

50

they were doing was just amazing.

51

I mean, it was, even today

would be amazing stuff.

52

And so I had no, you know, even

affiliation with computational

53

design or parametric design even.

54

So I realized that that was the

time, that was like a turning point

55

for me because I realized that

I was, I was behind the curve.

56

I, you know, I wasn't going to be designer

that I wanted to be without knowing at

57

least some of these tools and methods.

58

So, fast forward about 2010, um,

59

left TT, I came back in 2008.

60

Um, at that time, BIM was a big

deal, or at least started to be.

61

And so I, that, I had worked, I had

good experience with, you know, Revit.

62

I had, um, worked on, when I was at

Arup, I worked on the first, their

63

first Revit project, um, in 2005.

64

Actually, so it was early, early days.

65

And so when I came to TT, it

was a natural progression for me

66

to just get involved with that.

67

Um, sort of, uh, uh, both the

marketing of, of using those tools,

68

of how great it was going to be,

but also just in terms of, you know,

69

spreading it out to the company.

70

Um, and then I took my first with, uh,

um, you know, the other person that

71

was, was, was, was, formed the group

with me, which is Jonathan Schumacher.

72

Both of us took a Grasshopper class,

um, from the Mode Lab folks in Brooklyn.

73

And, you know, the rest

is history from there.

74

Um, that was like, you know, a

new language that we learned.

75

Um, uh, and he got great at it, I got,

I still am a beginner, effectively.

76

But at least, the point

is, I knew what it can do.

77

And I knew it could

hand back the business.

78

The

79

Randall Stevens: did that first,

uh, year, that first gathering,

80

did you all do it again?

81

Did you end up doing a series

of those or was that a one time?

82

Robert Otani: Um,

83

Randall Stevens: when you first had

the, that, that group got together

84

that first year for the computational

85

Robert Otani: Oh, no, that was it.

86

And, you know, and then, you know,

It opened up a whole new world,

87

actually, because in about 2010,

I think, Grasshopper was released.

88

Or, or, came out to the industry.

89

I was teaching at the time at Pratt

Institute as well, and the students

90

were using it, and I said, Oh, hope,

you know, that's something that

91

I think I can get my hand around.

92

It was very logical, um, you

know, it was very powerful.

93

And, you know, um, but, I didn't

even know Rhino at the time, so for

94

me it was a steep learning curve.

95

But, you know, after a while it was like,

this is, this is the, this is the, this

96

is, well, it's not the only way to go,

but definitely if you want to expand your,

97

um, capabilities, it's the way to go.

98

Randall Stevens: And were you

using those tools, uh, initially,

99

like in your workflows at TT or?

100

Yeah,

101

Robert Otani: so, um, in and around

that time, um, we were using, because

102

we were using Revit, because we were

using some, uh, structural analysis

103

tools, and we were using Rhino, um,

there was these connectors, and they're

104

still out there, called Geometry Gym.

105

Um, Yeah.

106

Yeah.

107

Um, and, um, we were using those tools

and I think the turning point was when

108

the CEO at the time, Tom Scarangelo, um,

one, he saw the power, but also the fact

109

that we're able to work in unison with the

architects who are also using Grasshopper.

110

We just had to make some modifications

to their scripts and we would get

111

the structural part and we would

work, you know, sort of, hand in

112

hand with what they were doing.

113

And we were doing things, you know,

in terms of getting documentation

114

for early designs, um, out there,

uh, wave, like, you know, orders of

115

magnitude, faster, cleaner, um, you

know, sort of more robust deliverables

116

than we would ever be able to do.

117

And so that, you know, that

changed the game for us.

118

Evan Troxel: you mentioned like you,

you had a steep learning curve if

119

you were to jump into this, right?

120

But, but just knowing what it could

do was super important, right?

121

Just kind of unlocking or just

understanding the potential there.

122

Robert Otani: Yeah.

123

Evan Troxel: I've taught leadership,

Grasshopper 2 leadership before

124

with that exact idea in mind.

125

It's not like I'm not trying to teach them

actually how to use Grasshopper, right?

126

I'm trying to teach them.

127

So that they understand.

128

I'm just trying to show them under

the hood so that they can understand

129

what's possible so that they can be the

enablers of the ones who are actually

130

going to be the operators of the tools.

131

And I'm just curious from your

standpoint, like what, how did

132

you take that information and how

did you deliver that to your CEO?

133

At the time to help them understand that

so that so that they would really let you

134

guys run with it and your team, right?

135

Because there's there is like

you you do want to match up the

136

architectural firms as well, right?

137

So you can you can deliver a higher level

of deliverable back to them in a faster

138

But but like efficiency is not the only

reason you You would do this, right?

139

And so I'm just wondering from a, you

know, a leadership standpoint and an

140

enablement standpoint, because there's

still firms out there just trying to

141

make this happen, this, this stuff that

you did 20 years ago or whatever was.

142

Right.

143

And, and this is still a challenge

in the industry for a lot of firms,

144

because, because there are so many high

tech tools that people use in firms

145

or can't use them because leadership

doesn't know what they can do and

146

they don't want anybody quote unquote,

wasting their time trying to learn these

147

things,

148

Robert Otani: your time.

149

Yeah.

150

So I think, I think it was a

combination of, um, um, you know,

151

at the time I was a pretty well

respected structural engineering,

152

structural engineer in the firm.

153

So it wasn't like I was blowing smoke.

154

I wasn't just the leader who, I

wasn't a leader at all, actually.

155

I was just kind of a, you know, mid

level, I would say, engineer at that time.

156

you know, but I did have a, uh, a good

reputation to working on complex projects.

157

Um, I think years later I was special

structures leader at Thornton Tomasetti.

158

So at least what I did say carried

some weight at least, but I think

159

the best, uh, feedback that we

received was from our clients.

160

Um, you know, Oh, this is cool shit.

161

You know, this is, this, you

know, this is, this is amazing.

162

And we

163

Evan Troxel: your clients

being the architects.

164

Robert Otani: Our clients

being the architects, yeah.

165

And we won a competition, won a project,

and you know, I would say if the only

166

thing you're going to get out of is good

marketing, well, that's directly to the

167

client that's going to be hiring you.

168

That's not the worst

thing in the world either.

169

Evan Troxel: right.

170

Robert Otani: the fact that we're able to

actually win a project was another bonus.

171

And the fact that you know,

we um, we're, we're giving a

172

better product was even better.

173

So, you know, it wasn't

a hard sell for me.

174

Cause I will say, you know, Tom

Scarangelo is, he was, he was always

175

a tech first, uh, sort of leader.

176

Uh, you know, he was the first person with

the black, with the BlackBerry, right?

177

Or the, no, even before the

BlackBerry, it was the PalmPilot.

178

Evan Troxel: Yeah

179

Robert Otani: You know, Tom had

the PalmPilot 4, the PalmPilot 7.

180

It was like, he always had the latest one.

181

And then, you know, the BlackBerrys

came around and then of course

182

the iPhones and all that.

183

So it wasn't, he, he

already had that mindset of.

184

You know, um, technology is business

and business is technology in many ways.

185

Evan Troxel: You're also describing

this very rational behavior

186

of a total engineer, right?

187

And, and I'm thinking of like

architects who are, they, we

188

think we know it all already and

there's nothing new under the sun.

189

And, and so it's, maybe it's a

different argument for a different,

190

uh, ear to actually hear it on an

architecture firm, but, but that's

191

the way you're describing it.

192

It's a no brainer, right?

193

It's like, I don't, I think a lot of

194

Randall Stevens: It's obvious.

195

Evan Troxel: with this.

196

Robert Otani: I mean, yeah,

I know what you're saying.

197

I will say, you know, I taught at an

architecture school for 10 years and

198

there is a different, you know, architects

are made differently in many ways.

199

But if you break it down into a business,

um, you know, Uh, use case, then,

200

then it becomes very logical that,

you know, it's, it's, it's marketing,

201

more projects, better workflow.

202

It's hard to argue

203

Evan Troxel: kind of ticks

all the boxes, right?

204

Yeah.

205

I

206

Robert Otani: it does tick all the boxes.

207

And now I will say tech, you know,

computational design is hard to scale.

208

So, that, that, that's where,

um, you know, um, you need

209

to do it at the right time.

210

It's not for every, you know,

point in time in a project.

211

And, um, uh, you know, I think that's

even now, that's still something

212

that we, we, we, we, we, we strive

to work on is because it's not taught

213

in college, at least for engineers,

architects actually are better prepared

214

than engineers by a lot, actually.

215

So, um, um, you know, we have

to sort of find the right

216

people with the right mindset.

217

who are going to leverage the

technology because it's one of those

218

things where uh, engineers like to

solve problems, that's what we do,

219

and so in many ways we're, when you

design parametrically, um, you're just

220

solving a problem in a different way, uh,

in a more data driven method as opposed

221

to, you know, an engineer likes to say,

I know the solution, this is what I'm

222

going to do, that's the right answer.

223

But the problem is that the

architect may not like that answer,

224

right?

225

So then you have to, then you

have to adjust your design.

226

So, um, you know, you know, I would say

it's still a relatively small percentage

227

of engineers who know, you know, how

to work that way, but it's changing.

228

It's getting better.

229

Yeah.

230

Randall Stevens: uh, maybe Bridges

or what was the evolution from you're

231

beginning to, you know, do this work

internally at TT and the evolution of

232

what became the CORE studio and kind

of more investment in these directions.

233

Robert Otani: well, as I mentioned, it

started with the, um, we had a, before

234

it was CORE Studio, it was a team

called Advanced Competitional Modeling.

235

Um, Very small team.

236

And, uh, when, and we started, um,

237

a, an R& D initiative in the firm.

238

And it was kind of, well, at the

time it felt only natural that, you

239

know, it would be at least funnel

through our team, the R& D part.

240

and I will say I have to give credit to

the, um, uh, the guys from AEC Hackathon.

241

I don't know if you remember those.

242

I don't know.

243

I think they still do one

Copenhagen, um, Hackathon.

244

Um, they did, I think it was their

first one, AEC Hackathon in, uh, the

245

Facebook headquarters in San Francisco.

246

And I sent four, three or

four of my team members there.

247

And, um, You know, they came,

they came back like kids

248

from a candy store, you know,

249

Yeah.

250

it was like, you know, they go out

there, first of all, some of them

251

probably haven't been to San Francisco,

so that was a, you know, fun by itself,

252

but then, they met all these like

minded people, they came back with,

253

Evan Troxel: It's a Burning Man.

254

It's like they found their tribe, right?

255

Robert Otani: yeah, yeah, so, I, we

still have some videos out, the funny

256

thing is, their, their hackathon

was to take an Excel spreadsheet And

257

move a dot from one point to another

point through, through, on the web.

258

That was it.

259

So, they learned the web.

260

That was 2013.

261

Evan Troxel: Wow.

262

Robert Otani: when they came back, again,

I'm not, I didn't go, but when I heard

263

all that, all that sort of commotion

about it, I said, why don't we do our own?

264

So that's, that's how we started our,

our, our conference and our hackathon,

265

was, was from that initial one.

266

And, um.

267

to this day, it's still like that

same feeling, you know, of, of finding

268

that, yeah, like a Burning Man.

269

It's kind of like the

AEC tech Burning Man.

270

Evan Troxel: Yeah.

271

Robert Otani: Um, and, and I, you know,

I, I think it's, it's a significant

272

event because it's sort of, anything

goes, you know, it's one of these

273

things that's not around any particular

technology or idea or company initiative.

274

It's very, um, you know.

275

Uh, you know, crowdsourced in many ways.

276

Evan Troxel: Yeah, you don't

direct the topics, right?

277

They from the other way around.

278

Yeah.

279

Robert Otani: And it's, I will say,

I, I even mentioned it at the end of

280

the hackathon, uh, a few weeks ago.

281

don't know how it happens.

282

So we have something called,

when we start the event, we have

283

something called the lightning round.

284

And anybody, again, most of the

people don't know each other.

285

So it's like, there's a hundred

people in a room, nobody knows each

286

other, and they just get up there

and say, Hey, I wish I could do this.

287

I don't know how to do it.

288

I wish I could do it.

289

In about a span of 15 minutes, 34 ideas

were, were discussed in the next 15 to

290

20 minutes, eight teams were formed.

291

I don't know how it happens.

292

It's like a psychological Petri dish

of, you know, behavioral Petri dish.

293

I don't know how it happens.

294

It's just organic.

295

It's just like people start going

in clusters and all away they go

296

for the next 20 something hours.

297

Randall Stevens: That's cool.

298

Evan Troxel: When you started the ACM

group, the Advanced Computational Modeling

299

group, it sounds very purpose built and it

300

sounds very much in service to

the design teams in your office.

301

But then the shift to opening

this up to R& D, which

302

is open ended, right?

303

Potentially.

304

Um, and then hackathons

and things like that.

305

How did you, did you, was there

a conscious shift or were these

306

just like, Oh, this feels right.

307

Or like how, if you could kind of put your

thumb on, on that, what was kind of the.

308

What was the trajectory

309

Robert Otani: Yeah.

310

I think each moment was a little

bit of a learning process for us.

311

So, you know, started with

computational modeling.

312

We saw how powerful it could be on

projects and how it, it, it was something

313

that if you're, uh, you're, uh, you know,

been around engineering for a while, it's

314

something you've wanted to do over the

years, but just, you know, the architect

315

says, I want to know that do this well, I

can only do this in this amount of time.

316

This is what you're

going to get, in a way.

317

And then you feel bad about it, because

you probably could have done that if you

318

had more time, or whatever other methods.

319

So, part of, um, that ACM group was

that we also did training, by the

320

way, um, as part of that initiative.

321

Because we didn't have projects,

you only, the interesting thing is

322

about computational modeling probably

only happens in about 2 percent of,

323

2 percent of a project timeline.

324

but it has a huge impact.

325

That was always the promise.

326

And that's, if you do it

right, that's how it works.

327

Um, maybe it's 5%.

328

Um, so we ended up training a lot

of people in that time period.

329

Um, not everyone's it's interesting

that people that gravitated to it

330

didn't always stick around company.

331

Um, they were the forward thinking

people and they started their

332

own companies or did some other

ones to do in other industries.

333

It was interesting that way, but.

334

That doesn't mean that

you don't do it, right?

335

That just means that

336

Randall Stevens: Sure.

337

Robert Otani: this, that was,

maybe this industry was not meant

338

for those, those people that are

gravitating towards advanced workflows.

339

Um, but the ones that did stay, you know,

became extremely, uh, valuable for us.

340

So the transition into R& D,

I think was more from the top.

341

Um, that was, I think.

342

you know, you have to think about what was

happening outside of our industry, which

343

was the Googles of the world and Facebooks

and Amazons, that there's something else

344

out there that our industry, you know,

is always kind of slow to, um, discover.

345

And so, um, you know, I would

say we always did some level of,

346

you know, project level R& D,

um, and try to move that forward.

347

But it was I think, you know, the CEO

and leadership at the time realized

348

you're not going to move the needle, um,

by, you know, trying to just scrape up

349

some good ideas and move them forward.

350

It has to be a holistic

concept within the firm.

351

and I think I'd also at a similar

time, um, we were, um, the company

352

that is, was, uh, Uh, having these,

every five years we have these five

353

year plans and you're, it's meant to

look, you know, pretty far out there.

354

Five years is a long time in our industry.

355

And so I think, you know, if you want

to be, again, our mantra, I think it

356

was developed around that time was

to be the global driver of change

357

and innovation in the industry.

358

Well, you don't do that, you know,

from taking, you know, Some cool

359

ideas from a project moving forward.

360

I think you have to invent, you know,

some real innovation, innovative ideas.

361

And so that's, I think when it became

362

I think, and I remember, I remember

the first, some of the first ideas

363

was, um, using drones in construction

and we actually use it on projects.

364

It was one of those things

where I'm like, Oh, why not do

365

inspections for using drones?

366

It was pretty simple.

367

but at that time, pretty novel.

368

another one was, which is still in

existence, it was this damper technology,

369

uh, that we've now called separate

company called Hummingbird that we, um,

370

someone came to us, I think it was from

NASA and said, Hey, you guys might be

371

interested in this damper technology that

we use for, you know, rocket launchers.

372

And we could apply that to do a building

and it's still, it's a separate company.

373

It's called Hummingbird Kinetics.

374

Um, so, um, so yeah, I think that was

a significant transition to formalize

375

innovation, if you will, and R& D, um,

Blue Sky R& D, you know, not something

376

that is, um, uh, you know, structured.

377

It's kind of unstructured R& D.

378

Randall Stevens: I'm assuming you

had to pull together an original

379

team and you got a budget approval

from the company to get the team.

380

What did that initial team look

like when you all started it?

381

Robert Otani: So, and Evan probably knows

this, you know, here in New York we had

382

this funnel of, uh, eight I would call

them A and E and software developers

383

from Stevens Institute of Technology.

384

Evan Troxel: Mm.

385

Robert Otani: Um, you know,

it was again, Nate Miller.

386

Um, I think, uh, I'm

trying to think a lot.

387

Again, I can't name all the names,

but a lot of people came from

388

that, that, that, that school.

389

And so we, even now we have a few here.

390

Um, actually my senior

director, Alex Pollack also.

391

Uh, was from Stevens.

392

We started getting those folks, um, and

they were an interesting group because

393

most of them were architects who went

to, uh, you know, this master's program

394

at Stevens Institute of Technology

called Product Architecture, which

395

was a combination of architecture,

engineering, and software, which

396

was exactly what we kind of wanted.

397

Um, you know, we, you know, we

wanted this sort of the mindset.

398

of an architect of solving hard

problems, um, the structure of an

399

engineering mind to, you know, um,

figure it out in a way and then the

400

software part to stitch it together.

401

And, um, you know, that worked out

really well for us, I would say.

402

Randall Stevens: Can you talk a little

bit, Rob, about, I know you all, you

403

all obviously develop some of your

own tools internally, but you also

404

team up with You know, other outside

software companies and do integrations.

405

One, how do you all think of,

you know, how, how do you think

406

about when do you build versus go,

you know, work with somebody or

407

help direct something externally?

408

And then, uh, maybe a couple of

examples where, where you all

409

have done that successfully.

410

And

411

Robert Otani: outsourcing.

412

One, because we, we've, that's in many

ways, when we do, it's sort of something

413

that we know we can do, but someone

can do it faster or better or cheaper.

414

And there's not that something

innovative within that task.

415

Um, if there is, then

we just, we, we do it.

416

Um, because, you know, if you hire

someone to do something that we may not

417

know the answer to, of how to do it.

418

Then it becomes a scope problem, you know,

additional change orders or whatever.

419

Randall Stevens: you lose that

learning that happened during

420

Robert Otani: Yeah, the fun, that's the

fun part for a software developer, right?

421

Is to learn something

inventive in that process.

422

Um, so we don't do a ton of it.

423

We have done it with our Konstru platform,

you know, which is a separate company now.

424

Um, The hard part about software is that,

particularly web development, is that

425

in five years you've got to rewrite it.

426

It's, you know, just things get old.

427

Um, and, you know, data structures change.

428

Uh, you know, so we've had to

rewrite a lot of that stuff.

429

Um, which is, which is, which is good.

430

It performs better.

431

It's, it's, it's, but it's

seamless for the user, end user.

432

Randall Stevens: So was that part of

y'all's mission, Rob, to, um, not only

433

develop things that would be used inside

TT, but, but commercialize those because,

434

you know, your mission of wanting to

change the industry, I would think that

435

that's, goes hand in hand with, it can't

just stay here if we really want to.

436

Robert Otani: I would say

the focus is internal.

437

But if something like Hummingbird,

like Konstru, comes out that we see

438

potentially a bigger benefit externally,

whether it's from, you know, from a

439

monetization standpoint, or whether

it's just not the exact fit, uh, for

440

Thornton Tomasetti as the engineering

firm, um, Then, yeah, there's always

441

an option to get it out there.

442

Hmm,

443

Randall Stevens: wants

to see it through and.

444

petitions to kind of spin it out.

445

Or what drives

446

Robert Otani: Again, I think

it's a business decision.

447

You know, um, Konstru, Uh, well, you

know, there was a series of companies.

448

So one of them was, uh, called T2D2.

449

AI, which is a computer vision tool that

does autonomous building inspections.

450

Um, You know, that one, we

do building inspections.

451

But if you really wanted to scale

that application, um, it, it

452

probably shouldn't be internal.

453

It should be external.

454

Um, I think the people that decided

to spin it out saw a huge potential

455

in where, where that could go.

456

Um, I will say this, that one of the

lessons that we learned was, um, a lot

457

of the technology that we've developed

has been ahead of the curve, um, which

458

is not always great for software, and

I'm sure you know that too as well.

459

From your

460

Randall Stevens: If you're

educating the market, right.

461

It's

462

Robert Otani: Yeah, um, you kind

of want, you kind of want to

463

build something that people need

at that particular point in time.

464

Um, but that doesn't mean that

it was the wrong thing to do.

465

It just means that, you know, it was

solving a Thornton Tomasetti problem.

466

Um, may not solve everyone

else's issues at the time.

467

Randall Stevens: you know, AI

obviously carries lots of buzz.

468

There's every, everybody's

got it somewhere says that

469

they've got it somewhere.

470

But maybe you can describe a couple places

where, where you are implementing or

471

using machine learning, um, and, you know,

just, you know, You know, where, where

472

do you see that going and how effective

473

it's going to be changed some of

the things that you are working on.

474

Robert Otani: Well, you know, maybe I'll

tell you the story of how I got behind it.

475

Um, it was around 2015.

476

Uh, we had a relatively small development

team and one of the developers, um, uh,

477

on the team, um, would always talk about

machine learning and I was like, what's

478

this machine learning all about, you know?

479

and then when he told me about, you

know, the fact, you know, you take

480

some smart data, Use some algorithms,

data science to train that, train

481

something that knows everything

about that, what's in that data set.

482

And then sooner or later, it gets

good enough, you just ask the question

483

and it gives you an answer based on

those previously solved problems.

484

And I was thinking to myself, isn't

that what we do as engineers and

485

experienced

486

Evan Troxel: that engineering?

487

Robert Otani: That's engineering, but

that's architecture, that's anything,

488

anything, anything as a human, right, over

489

Evan Troxel: No, no, no, not architecture,

490

Rob,

491

Robert Otani: architecture,

parts of architecture, I should

492

say parts of architecture.

493

Um, so yes, engineering is more specific

in that, in that, in that realm.

494

But, um, so that, that got me excited

because that, that means if you do it

495

right, um, it becomes a significant um,

uh, sort of tool for knowledge transfer.

496

Like just knowledge.

497

Um, even if it's just an application,

it's like taking someone's best, the

498

company's best Excel, uh, tools, instead

of having everyone has their own, right?

499

Aggregate them.

500

Let's create one or two, say, and

then everyone has the same thing.

501

We're not rebuilding them

over the ages of time.

502

Um, the other aspect of that.

503

Yeah.

504

that I saw was, and I, I was at

a conference last week and um,

505

and it sort of dawned on me.

506

It was, you know, probably six, maybe

more, 70 to 80 percent of our project,

507

every project that we do has been done

before, like literally done before.

508

Whether it's, if you break it down

into its components, whether it's

509

a foundation, a column, a beam, or

slab core, it's been done before.

510

So why are we spending months, years

sometimes, redoing the same thing?

511

Um, almost the same thing, I would say.

512

Um, and that's, that's, that's the

promise of machine learning, is that

513

you aggregate those solutions, um, to

give recommendations for a particular

514

project when they need it, um, as

opposed to reinventing the wheel every

515

time, which is what we do, actually.

516

If, if, look, if, if someone asked

me about a project we did ten years

517

ago, maybe I can ask five people

that may have worked on that project,

518

their, their, their memories might

be fuzzy, but we don't even do that.

519

Um, So there's no sort of, you know,

so in many ways, so what's the value of

520

a company if you can't, you know, take

past experiences and move them at the

521

product when you need it at the time.

522

So, you know, that's the big picture.

523

Um, the smaller picture what we're doing

is just creating these little micro

524

apps that will save some time, you know,

during particularly schematic design.

525

Um, and you know, I, and on the other

thing, the correlation that I made,

526

I made it in that presentation, um,

was What we did when I first started,

527

which was in the mid nineties, we

didn't have computers at the desk.

528

Um, so we had to do

everything by not all by hand.

529

It was, we had shared computers, so

we would do like Excel spreadsheets

530

over there, or, um, you know, we

didn't want to add, well, there was

531

a few engineering apps, you know, um,

DOS programs and things like that.

532

Um, but the rest of it was

very much person to person.

533

And you would use past project

data for what you were doing.

534

So we had, and I still have it actually

here back in my pile of mess back there.

535

Um, all these little, um, single

bay design apps for office buildings

536

and for other types of buildings.

537

And so, you know, an engineer

would do one, they'd do it at nice.

538

And we had CAD at the time,

so you can do it in CAD.

539

We had an army of CAD drafters back then,

way more technical, like drafting staff

540

than relative to what we have today.

541

Because engineers and

architects just do it now.

542

But, um, the point is, is that once

we did it, you never did it again.

543

You go to the book, you know.

544

And we, a good engineer can

design a building without

545

doing any, almost any analysis.

546

But, so in the last 15 years,

the tools are very good.

547

Uh, you know, from drafting tools

or BIM tools to analysis tools.

548

To the point where they can do the entire

building from every part of the building.

549

Um, in, you know, probably

days or a week or two.

550

But the point is, is that once you do it,

it's very hard to change those things.

551

And, arguably, the designs are not that

much better, um, because what you're

552

doing in those early phases is just

constantly, I would say, responding

553

to, you know, potential modifications.

554

You know, it could be a cost issue,

it could be a coordination issue, it

555

could be a complete design change.

556

Um, and so, you know, why do you

need to be designing a whole building

557

when you know, the 50 percent of

the building is going to change.

558

And so that's, that's sort of the,

um, what I use as a corollary to

559

what we did when I first started,

which is a very data driven approach,

560

um, to what machine learning can

do, um, you know, in the future,

561

Evan Troxel: From kind of a disruptive

standpoint with software and with machine

562

learning and like encoding knowledge, was

it really seen as disruptive, potentially,

563

to the future of an engineer's Workload.

564

Um, I mean, I'm just curious how

you guys kind of saw that or how you

565

pushed through that in your offices.

566

Robert Otani: though we're still

pushing through it, to be honest.

567

But what it is, is, um, engineers

are taught to be, to, um, I wouldn't

568

say wary, but, you know, question

what's put in front of them.

569

In fact.

570

We generally tell engineers that they

need to Excel, even with someone gives

571

them an Excel spreadsheet to check it.

572

And so, machine learning is very

black box, you don't see any of that.

573

It's, put some stuff in there

and something comes out.

574

Um, so, you know, what we've been

saying is, okay, we have the main,

575

we made this app that does this, you

should use the app only after you've,

576

you know, um, you know, done it.

577

in your own methods or conventional,

what I call conventional methods,

578

soft, uh, commercial software methods.

579

So you can get comfortable with the,

in the level of, of, of accuracy.

580

Once you get there, then you don't need

to push those a hundred, 200 buttons,

581

you know, uh, to get that answer.

582

It's the answers there.

583

Evan Troxel: Mm hmm.

584

Robert Otani: Um, even better is when

we get ideas for creating another

585

app, you know, that's, that's sort

of, we kind of want to create this.

586

Culture of, um, you know, continuously

improving the apps that we have and

587

then suggesting new apps for the future.

588

Um, we know in our team that

we don't have all the answers.

589

It's, it's, you know, we're relying on

some, you know, the, the, the, the most

590

widely used types of tools or use cases.

591

And then, um, there's always something

that, um, you know, either, sometimes

592

it's different region, different

regions do things differently, or,

593

so there's going to be a lot of

flavors, I think, over time, even

594

doing the same thing, same apps.

595

Randall Stevens: Rob, when we did

the, uh, we did the confluence event

596

in Brooklyn back in April, you know,

you, you sent a couple of people from

597

your team, Omid and Sergei came and

598

presented one of the, they spent

some time, you know, telling, uh,

599

explaining the story of, uh, I guess

the gentleman's name was Mike that

600

you all.

601

You all did a project where somebody

that was longtime, uh, leader inside

602

the firm, uh, passed and, uh, but

you were able to, uh, drive a project

603

that was like a collection of a lot of

information that he had left behind.

604

Can you maybe tell us, tell us that

story and, and where that stands?

605

I loved, uh, this whole idea of like

capturing, you said it earlier, capturing.

606

knowledge, and then having

that live on is a, has a,

607

lot of appeal.

608

Robert Otani: well, I'll give

you, I'll tell you the story, and

609

I will say it's not going to be.

610

easy for everyone, but it was easy

for Mike's, um, um, conversations

611

and data because of how good he was.

612

But, um, so back in 2022

613

or something like that, one of our, Omid

actually had an R& D project with, with,

614

with the R& D program to train a model,

to learn the building code that was, or

615

at least find things in the building code.

616

And, um, you know, we, we're going

to create something custom, you

617

know, and, and, you know, and all

of a sudden an open AI comes out.

618

So we, he created his own little web

app that, you know, uses, you know, use

619

the open AI API, um, foundational model.

620

Uh, points to indexes, you know.

621

I think we actually did

index a building code.

622

Um, I don't do that full time because

I don't know the legality of that.

623

I think, But we just did it

as a test case, and it was, it

624

was amazing what it could do.

625

You know, it basically reads through a

PDF in seconds, less than, you know, and

626

then gives you an answer based on that.

627

Um, we also did some other AI, which

was to sort of explode, if you will.

628

All the tables and charts

which are frozen in PDFs.

629

Um, so they're extracting the,

the data that's within those

630

tables that are generally frozen.

631

So it was really good.

632

I mean it was finding

everything that we wanted.

633

So, um, and we had different use cases.

634

We're, we're indexing our

intranet, which we, we call Spark.

635

Um, it's sort of the internal sort

of, you know, intranet of, we post,

636

you know, technical information,

resources, but also just fun things.

637

And, um, then it showed, you know,

a lot of promise for that because it

638

was finding things at the internal

search engine to that application,

639

which is absolutely terrible.

640

It was finding things not just

from the beginning of time, but

641

also in chronological order,

which also is very valuable.

642

Um, so Mike heard about it and he said,

Hey, you know, I got, you know, 10 years

643

or so, or I think it was 7, I forget how

many years it was, of um, emails, because

644

in the last, I think it was 7 years,

because in the last 7 years, he wasn't

645

actually working on particular projects,

he was just an internal resource.

646

Um, he used to be our sort of, uh,

site, um, you know, um, engineer, just

647

to, you know, solve all the problems

before they became real problems.

648

And he was just so knowledgeable

about primarily steel, but pretty

649

much anything that happened on

the site as related to structures.

650

And so in the last seven

years, he was just a resource.

651

Anybody can ask him a question about,

you know, steel or anything on the site.

652

And he saved every

single email judiciously.

653

And the other part of it is when

he answered those questions, He

654

didn't just give them the answer.

655

He said, based on ASTM, blah,

blah, blah, and AISC, this, that.

656

So it was very, like,

robotic in a way, right?

657

Of, you know, almost like a Turing, you

know, sort of response, if you will.

658

And, um, so okay, we said,

decided to take that.

659

He didn't know how to do it, but,

um, so he asked me to do that.

660

So we took his email,

cleaned it up a little bit.

661

Because you don't want

all that stuff to date.

662

Well, maybe you want to

date, but we cleaned it up.

663

And we had him create 30 questions.

664

Like, tricky questions that,

you know, not everyday stuff.

665

and when OpenAI responded to

those questions based on that data

666

set, um, we asked him to rate it.

667

And he gave it a 4.

668

7 out of 5.

669

Which You know, as I, yeah,

it's very, it's very, in a way

670

that's very data driven as well.

671

Right.

672

Evan Troxel: Right.

673

Robert Otani: Um, but that was

the only method that we, we can

674

think of to say, okay, this is

good or this is bad and his 4.

675

7 is like, you know, it's, it's,

it's way better than anyone else's 4.

676

7.

677

So, so even now we have, and

unfortunately six months after

678

that, he passed and you know, so.

679

Um, again, we, we, we actually, so,

you know, our ex CEO and a couple of

680

others, uh, went to a funeral and,

uh, we did, um, start or assist in

681

creating a scholarship in his name

at his university that he went to.

682

Um, and, um, his name is Michael Ashmit,

by the way, and, um, We decided to

683

take that data set, we got permission

from the, from his family members, to

684

use it, you know, for, for the rest of

eternity, and we still have that sort

685

of, uh, you know, data set, if you will.

686

Evan Troxel: Yeah, in a

way he's living on, right?

687

Like this, that's

688

pretty incredible resource.

689

And like you said, it's not easy

for firms to, because not everybody

690

did what he did does what he did,

691

um, which I think is, is a key

part of that too, as a, as a good

692

disclaimer for you to make up front

because yeah, a lot of people.

693

Don't go through that rigorous

process of cataloging that information

694

along the way, but that's kind of

what you need to build a mic tool,

695

right?

696

Randall Stevens: So was he

a teacher at heart, Rob?

697

So

698

Robert Otani: Oh yeah.

699

Oh

700

yeah, he, you know, he was so

smart, you know, very, very smart

701

people sometimes can get short

and, you know, give short answers.

702

Um, but he was not like that.

703

He would

704

Evan Troxel: he had citations and

705

Robert Otani: yes, like patients.

706

Yeah.

707

Evan Troxel: That's

708

Robert Otani: I mean, you know, so

not, you know, there's, there's only a

709

few Mike's out there, there's a couple

issues there that, um, that if you think

710

about it is not one, it's not creepy.

711

because, you know, that's the first

people think, Oh, it's creepy to use

712

someone else's past, but it's not

because that's what we do all the time.

713

You know, when, when we have a

conversation, we work with someone for

714

a long period of time, it's inevitable

that that person's sort of learnings

715

and, and, and teachings, I should say.

716

Um, and you know, you don't know what's

going to come through you, but it does.

717

Um, you know, I still remember things

that Charlie Thornton said, you

718

know, Richard, Richard to actually.

719

That, that brings to the next point,

which was, if we can do that now,

720

then it makes it even more important

to catalog knowledge in a way.

721

Randall Stevens: Sure.

722

Robert Otani: Um, so,

723

Randall Stevens: Well,

724

Robert Otani: not going to

capture everything, like, you

725

know, it's, it's impossible.

726

Randall Stevens: well, I think, uh, you

know, part of that discussion that we

727

had back in April with Omid and, uh,

after they gave the presentation and we

728

kind of opened it up for conversation,

it was, One thing about machine learning

729

and AI is usually you're thinking about

that being on the data that you're going

730

to learn from is usually an end result.

731

And I thought it was very

intriguing to think about that.

732

What you're missing from end result is

the process that led to the end result.

733

The why, the why, right?

734

Those, that's harder to capture, but

you know, that, Those emails and that

735

explanation, you know, is how you learn

and, you know, in today's environment,

736

when you get, you know, especially if

young people aren't even coming into

737

an office to be mentored, you know,

738

Robert Otani: Oh, it's a, it's a, it's, I

739

Randall Stevens: It's a

whole, whole new world, right?

740

About how are people going to learn?

741

Why?

742

And it's always through stories

and all this stuff has passed and

743

absorbed over time.

744

Evan Troxel: listening to your

supervisor yell at somebody over the

745

Randall Stevens: Yeah.

746

Evan Troxel: like like

747

Randall Stevens: Well, I was, uh, I'd

given the example when we were at the

748

Confluence event talking about this.

749

I was like, I'd had the same conversation.

750

You know, I have a degree in

architecture and been involved with

751

the school here locally over the years.

752

And I was arguing, I was making argument

about like, that the real value of the

753

education is probably the desk crit.

754

So it's like, Why don't we

drop a, you know, drop a

755

recording device, an orb, right?

756

That's like,

757

it's red.

758

So everybody knows it's there, but

you're going to like record this.

759

And it's like capturing that conversation

between the student and the professor in

760

the studio time, to me is like, that's

where the actual learning happened, but

761

it, you know, at each desk, it's only

20 minutes at a time, but if you could

762

take and capture all of that and, and,

763

and put it into AI, you know,

engine like this, to me,

764

that's like a very.

765

Intriguing, sexy kind of thing to

think about happening, you know?

766

Robert Otani: yeah, and it may

not have value until you have,

767

you know, 50 desk crits, right?

768

Um, but it, it, there will

be value there, right?

769

It, it, and, um, you know, I was

thinking about it was, what's actually

770

hard about machine learning or, or,

or, or building new models, let's

771

just say, is that, you know, human,

like the, When I explain that, Mike

772

very clearly explained his rationale.

773

Um, humans do a terrible

job of that, right?

774

They like, sometimes

they get to an answer.

775

They don't exactly know

how they got to an answer.

776

So, you know, how you decompose

a response to some, a series of

777

problems in a way, your mind has the

most powerful parametric, you know,

778

model modeling software in the world.

779

And it's doing all these calculations

based on past experience, past

780

failures, past wins, all these things.

781

And you come to an answer.

782

You don't exactly know how.

783

And so that's why, particularly, you know,

to your point, Evan, about architecture,

784

it's much harder to train models on

architecture because it could be, look, it

785

could be a look from a client, you know,

or it could be a look from someone in

786

the, who's, who's, who's judging at work.

787

Has nothing to do with logic or had

nothing to do with the design, actually.

788

Yeah.

789

So that's very hard.

790

Um, and that's why, that's why I

don't think AI is going to take

791

over everything that we do because

there's, there's a human element in

792

there that, uh, it calls it illogic,

illogical behavior that is just going to

793

Evan Troxel: Yeah.

794

Emotional.

795

Robert Otani: so anything that we do

now, architects are doing, you know,

796

early design studies using past imagery.

797

It's kind of like precedent study and,

you know, automation, if you will.

798

Um, but it, It's really good, you know,

um, and, and it saves some time, and it

799

potentially sometimes spurs new ideas,

so why not do, do those types of things,

800

Randall Stevens: Yeah.

801

It seems pretty obvious that to me, that

it's going to be a companion, right?

802

It's

803

Robert Otani: Yeah.

804

Randall Stevens: it's going to help

either, like you said, accelerate

805

how quickly you can get to

806

an answer, or it's going to

spur you in some new directions.

807

Robert Otani: mean, if you think

about even the tech companies

808

now, right, whether it's autopilot

or, you know, self driving cars.

809

Um, or, uh, even what, you know,

Amazon does or Instagram, this,

810

that's easy stuff, if you will.

811

It's like, if this, then that, then it's

this, you know, then here's the answer.

812

What we do in our business

is way harder than that.

813

And, you know, look,

recently, you know, Zillow.

814

com tried to automate, you know,

real estate, it failed miserably

815

because real estate's not just a

series of boxes and numbers, it's.

816

There's, you know, people have a,

depending on who that person is, they're

817

going to have a different idea about what

some value of a house may be, or sometimes

818

it's just the salesperson is a better, can

pitch a sale better than another person.

819

And they lost their shirt, right?

820

They had to quit that.

821

Randall Stevens: they were trying to do

arbitrage on buying and selling, right?

822

And they lost, you know,

they lost that bet.

823

Robert Otani: So that's because there

was too many factors in that use case

824

that, and they didn't have the data

that, you know, driving those answers.

825

Um, so, you know, I think, I think,

you know, the idea with AI, at least

826

for now, is to focus on the things that

it can do well, and then we figure out

827

how, you know, how far we can push it

828

Randall Stevens: Yeah.

829

Robert Otani: down the road.

830

Randall Stevens: you've got quite the

team now that makes up the CORE studio,

831

so maybe you can give people kind of

a sense of scale, the team, and then,

832

uh, Maybe just a glimpse of what you

are working on now or what you think is

833

Robert Otani: Yeah, it's

about 40 people now.

834

The majority, I shouldn't

say the majority, but half of

835

them are software developers.

836

Um, we have, we have, so I'll just,

I'll, I'll mention the verticals.

837

Um, so we have an app dev team, um,

a CORE modeling team, which is kind

838

of like the continuation of the

advanced computational modeling team.

839

Um, data and knowledge.

840

So we have a team that's sort of cleaning

our corporate, you know, information, you

841

know, data, both data analytics and AI.

842

Um, a BIM team and something

that we call CORE delivery, which

843

is more on the fabrication, uh,

uh, design and fabrication side,

844

Tekla and things like that.

845

Um, so yeah, I mean, I think the big

initiative for that, for that, that

846

team is to make technology just part

of our everyday process, not something

847

that is, And a bolt on, if you will.

848

It's like what we want to do is the

day they walk in the door, they get

849

introduced to it, trained on it, and

then it just becomes part of the process.

850

Just, just a normal delivery.

851

We, and if we know that if they do that,

we can achieve that, that, you know,

852

we'll have a, a high performing team.

853

oh, I should, I should mention

we have the AI team, which is

854

really part of the app dev.

855

team.

856

Um, so that's the other part.

857

I would say the reason why I

always, I sometimes miss that is

858

because, um, the AI team and the

app dev team work hand in hand.

859

Um, you know, you can trade all the

machine learning models you want,

860

but if you can't spin it up into

something that people can use, it's

861

not, it doesn't do anybody any good.

862

So, uh, we've built all kinds

of tools around, um, just the

863

infrastructure of releasing machine

learning models and storing models

864

and, experimenting with models.

865

so yeah, I think it's, we, we've

gotten big enough where, um, we have

866

to start to prioritize now, um, with

what we do, because, you know, if

867

I had my druthers, I'd, we'd be 40

people, you know, but we're not, so

868

we have to focus on, you know, effort

and impact with an eye on the future.

869

I think that's, that's critical,

because if you just focus on effort

870

and impact, you're going to solve

the next one or two years, but not

871

forget about the next five or 10.

872

So we have to kind of continue

to, um, to, to do that.

873

And which is why we do the conferences,

because I think that's a good way to

874

reset and see where we are and maybe we

have to, you know, change direction a

875

bit based on where the industry is going.

876

Randall Stevens: I'm

877

Robert Otani: why I go to events

878

Randall Stevens: I'm sure it's helpful

for recruiting too, right, just

879

to, again, find the people that are

880

Robert Otani: Absolutely.

881

Yeah.

882

Yeah.

883

Randall Stevens: Yeah.

884

I was just going to say, you know, uh,

it was, it's, you know, First of all,

885

thanks again for participating in our,

in our Confluence event this past fall.

886

It was great having you here.

887

We had a good, I think

a good group of people,

888

uh, some good presentations.

889

Yep.

890

And then, uh, and then, you know,

thanks for putting on the, you know,

891

doing the hackathons and I think it is

important to get, uh, these like minded

892

people excited and out and thinking

893

about the things and working on them.

894

Uh, it's, it's important

part for the industry.

895

And, um, and, uh, Obviously, Thornton

Tomasetti's a structural engineering

896

firm, but obviously the, uh, works hand

in hand with lots of, uh, the architecture

897

firms that Evan and I are closer to.

898

So,

899

Robert Otani: Yeah, if I were to say,

you know, one thing that I think we need

900

to do a better job of, not just us, but

across the sort of, you know, technology

901

spaces, creating apps that benefit all of

us, um, as opposed to just the engineering

902

firm or just the architecture firm.

903

I think if we start to create apps that,

you know, impact the entire process of

904

design and delivery, I think, um, then

we'll start to make some real impact.

905

you know, in the, in the industry.

906

Evan Troxel: Yeah, I think the first

time that I became aware of what you

907

were doing was when, I can't remember

who it was exactly, it might have been

908

you, but at one of the USC BIMBOPS,

we talked about Asterisk, and it being

909

this, this web app, right, that you

could throw a rhino massing model at

910

it, and it would spit, you could say,

Mass timber, concrete, steel, right?

911

And, and it was, it would

give you information back.

912

And as a designer, right?

913

As an architect, it would be okay.

914

I'm way ahead of where I would

have been without this tool.

915

And you're sharing that, that

ability with me to do that.

916

And then it enhances what I'm doing.

917

But if it, if it comes back to you

as a next, as a step in the process.

918

Your way ahead of the curve as well.

919

Right.

920

And so just to really speak to what

you just said about going outside

921

of your own silo and creating

useful software for the industry.

922

I mean, that speaks to that directly.

923

So

924

that's really

925

Robert Otani: Yeah.

926

You know, um, that, that, by

the way, that app still works.

927

Um, and that was, I, I still

think it's an amazing app, but

928

it was too early, too soon.

929

Um, engineers were like, holy

shit, this is totally black box.

930

I have no idea what's happening.

931

Um,

932

no, I can't trust it, but it

was doing, you know, things.

933

3 person weeks in about a minute.

934

Evan Troxel: Yeah.

935

Robert Otani: it was doing.

936

I showed that in a presentation as well.

937

We have a newer version of that.

938

But, that application would be

used for like 2 hours in a project.

939

That's it.

940

but you would, to your point, you

would be that much smarter, that

941

much further ahead, and we wouldn't

have to churn through 10 iterations.

942

Um, That's the idea.

943

And you know, that's, it's one of these

things where like, you know, sometimes

944

you have to disrupt yourself, um,

you know, to, to make an improvement.

945

If we stick with exactly what we've

done year after year after year, we're

946

never gonna improve, um, you know, how

people communicate with us or how the

947

project goes so that, that was asterisk

in many ways is, is a disruptor.

948

Um, but it's a good one, I think.

949

I think it's more of what the team needs.

950

You can think about it.

951

Imagine if we had all the other things.

952

Mechanical, electrical, plumbing.

953

I mean, it would be

pretty amazing, you know?

954

And architecture, for that matter.

955

Randall Stevens: Great.

956

Well, thanks again, Rob.

957

This has been a great conversation and

it's always, uh, fun to watch, you know,

958

not only what you are doing in the studio,

but obviously hosting these events and

959

getting others out and, uh, young people

interested in, uh, what you're doing.

960

You know, everybody coming out of school

now, like you said, a lot of them are

961

already using Rhino, and they're using

962

Grasshopper, and Revit, and Dynamo,

and all those types of things, so

963

there's at least, uh, some, the

audience is a little prepped as they

964

come out, a little more than they were,

you know, 10 or 15 years ago, but,

965

uh, yeah, it's been

great having you join us.

966

We

967

Robert Otani: Well, thanks for having me.