Carl Veillette, Chief Product Officer at Newforma, joins Evan Troxel and Randall Stevens to talk about the data problem hiding underneath every project. Anchored by the new Confluence Tech Stack Survey, where the tool count per project jumped from about 15 a few years ago to 84, Veillette argues the real issue is not the number of tools but the broken connections between them. He makes the case that a single source of truth does not exist, and that the goal is a single view across systems like Teams, Outlook, SharePoint, ACC, ProjectWise, and Procore. They get concrete about the golden thread, governance and retention, and why Newforma's job is to free the data from the files rather than host it.

This episode is especially relevant for CIOs, digital practice leaders, and project information managers wrestling with data sprawl, litigation risk, and AI adoption. Veillette walks through practical AI use cases, from mining the inbox for project records to agentic workflows for code compliance and lessons learned. The throughline is blunt and useful: AI is only as good as the data behind it, so the value lives in the record, not the model.

Episode Links:

Connect with the guest

• Carl Veillette, Chief Product Officer, Newforma; co-founder of BIM Track

The survey

• Confluence Tech Stack Survey (2025) — the report behind the 15-to-84 tool sprawl discussion

Tools and systems mentioned

• Microsoft Teams, Procore, Autodesk Construction Cloud, Bentley ProjectWise, Egnyte, Panzura, Miro

AI and code compliance

• UpCodes, Archistar, Workorb AI, Knowledge Architecture

Standards and regulation

• ISO 19650, ISO 9001, UK Building Safety Act 2022

Watch this episode on YouTube.

-----

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 TRXL with me as usual.

3

And we've got a, maybe

our first repeat guest.

4

I can't remember if we've

had, uh, had somebody on

5

Carl Veillette: a big title.

6

Evan Troxel: so.

7

But, but if you're, but if you're

wrong, I mean, we're bo we're

8

Randall Stevens: I know.

9

Well,

10

Evan Troxel: so,

11

Randall Stevens: can fact check, right?

12

It's like the, uh, the uh, on, uh,

on x, uh, whatever community note.

13

Somebody can make a

community note about it.

14

But, uh, anyway, we've got, uh, Carl

Vati from, uh, new former with us.

15

Uh, I was looking back 'cause

I was reviewing, uh, you know,

16

the, the last conversation we

had and it was almost a year ago.

17

So, and I was joking just before we

came on here that I'm a year older and

18

my, I've started to let my hair dry

'cause I had it, uh, cut pretty short

19

on that last, uh, on that last call.

20

But, uh, glad to have you back on Carl.

21

And, you know, one of the things I think

about, uh, you know, we, we literally have

22

about a year under our belt from the last

time we talked, but things are moving.

23

You know, pretty quickly with

these kind of technology shifts.

24

So I think it's, uh, appropriate for us to

kind of jump back on and, uh, 'cause last

25

year we were beginning to talk about how's

the AI gonna affect some of these things.

26

I know you guys have been working

on some of these strategies.

27

Add new form and then also kind of

topical was that we just put out

28

recently this, uh, confluence, uh, tech

stack survey, technology stack survey.

29

So I know you've had a little

bit of time to look at that.

30

It's kind of hot off, fresh off the

presses, but there's a lot of things

31

I think that are there's, that are in

that report that, you know, kind of sets

32

the stage for just this complexity of

the ecosystem, which is a lot of what

33

we were talking about a year ago too.

34

Um, so anyway, but, uh, welcome,

welcome back on and, uh, maybe, uh,

35

maybe we can kick, kick this off.

36

Um.

37

I, I, you know, I was reviewing some

of the stuff that you've been, uh,

38

talking about over the last year with

what you're doing in new form, and one

39

of those things is you kind of refer

to this information sprawl and, uh, and

40

how they kind of get this golden thread

that can kind of weave back through.

41

I always, I, I talk about the same thing

with a veil, that there's this thread

42

that, uh, we have a feature called related

content that's like, lets you connect this

43

information, but maybe we can kind of kick

off there and, uh, maybe explain what you

44

mean by that golden thread and then we

can, uh, we can kind of dig into maybe,

45

uh, maybe this, this application sprawl

and just where all this information is.

46

Carl Veillette: Yeah.

47

So I, I think, uh, you know, when I

think about, um, our tech stack these

48

days, and I guess that relates to the

survey that you guys did recently, right?

49

But, um, you know, the, the tech stack

is growing and there's, um, you know,

50

there's a survey we did a few years back

and we asked the question, how many tools

51

do you have in your toolbox on a project?

52

Right?

53

And the answer was 15, right?

54

At the time.

55

And, uh, what struck me in your

survey is the number now is 84, right?

56

So that's 84.

57

The number of tools that organizations

are, are leveraging, right?

58

So.

59

I think this is like a, you know, this

is great because people are taking

60

the best tool for the task and the

activities that they have to perform.

61

Right?

62

Um, so, um, this is great

for the industry, right?

63

To, to be able to, you know,

have more technology choices and

64

best in class tools for specific

things that they are trying to do.

65

But at the same time, it is creating a

bit of, um, uh, governance issues, right?

66

When it comes to data, right?

67

And, uh, I think what we've seen in

the last year is, um, this increase

68

in amount of tools and storage silos

and, you know, it creates datas for

69

all issues and it's preventing you from

gathering insights and things like that.

70

And think about the golden tread, right?

71

In the uk it's a big thing because

of the creation is in place, right?

72

And, um, this is, uh, this

is also contributing to, um.

73

You know, that data scroll

and those information silos to

74

break the golden thread, right?

75

I think about the golden thread of

information as the, the ability to

76

go back, you know, maybe like 15,

10 years down the road when the

77

project's done and be able to reconcile

the audit trail of what happened

78

when we sent what to who, right?

79

And the ability to understand the

context of the project at the time.

80

Right?

81

So for me, like a golden tread is

not necessarily a single source

82

of truth because this thing

in my opinion, doesn't exist.

83

Uh, there's multiple

source of truth, right?

84

And it's all about trying to

connect the dots together, right?

85

And I think, you know, a lot

of people talk about command

86

environment and those sort of things.

87

Uh, some vendors call themself a

command time environment, which I

88

think doesn't really exist, right?

89

It's an a series of interconnected

tools that exchange information.

90

So, um, really like the mission

we've been about, um, in the last,

91

uh, you know, couple of years.

92

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

a mission that's, uh, that's even

93

more difficult these days now that

there are a lot more tools, right?

94

Is, is to connect the dots so that

we can create a single view of

95

information as opposed to a single

source of information, right?

96

So, you know, when it comes to

technology adoption, I think one

97

of the biggest challenge is, um,

well change management, right?

98

Because we're all humans and, uh,

we like to work certain way, we

99

don't like to change things, right?

100

Um, so, you know, really where we've

been focused is, uh, not trying to

101

change the way people work, right?

102

Uh, but also create kind of an information

layer that connects to those various

103

best in class tools that people use.

104

So you think about, um,

Microsoft teams, right?

105

So it's unlikely you're gonna displace

Microsoft team these days because it's a

106

communication channel, number one, right?

107

Uh, for, for project teams, right?

108

And.

109

Uh, I guess I would argue people thought

they, they were gonna replace emails, you

110

know, you know, a few years ago, right?

111

And this never happened.

112

It's not gonna happen.

113

Uh, we look at the statistics, right?

114

Number of emails increase on projects

these days as a result of more complex

115

projects and bigger teams and, uh,

regulations and, and such, right?

116

So I guess the data proves

everybody wrong on this.

117

You know, I was the first to think we

could replace emails in the been track

118

days, but, you know, this hasn't happened.

119

So now we just, we just have to, um, you

know, uh, connect those things together.

120

And that's the approach we, we've, we've

been taking is like, okay, there is gonna

121

be instant messaging in Microsoft teams.

122

There's gonna be meeting minutes

captured with AI in there, there's

123

gonna be emails, uh, so those are

all communications related, right?

124

But you're gonna have like

a flow of information.

125

Maybe a designer works with a, a GC

that's working out of Procore, right?

126

So you've, you've, you're, you've

got all those, those information

127

coming from various sources, right?

128

So.

129

Um, in instead of replacing those

sources or trying to be the center

130

of the universe, um, why don't

we connect those things together?

131

And then why don't we, um, create a,

a another trail kind of layer, uh, a

132

data connection system that integrate

with those tools and, and bring the

133

information in the context of the project.

134

So when the last phase of construction

project kicks in, I'm talking

135

about litigation here, you've got

the compete project record, right?

136

You've got the evidences, you've got

the information that's been shared.

137

Um, so for, for us, it, it's really

been looking at, um, you know,

138

where is the data flowing from

and, um, kind of connecting that

139

with the project activities, right?

140

So, uh, for, for, for us, you know,

the, the golden thread is, is that

141

ability to look at the complete project

archive, right from the, the very

142

early stage of the project, all the

way to delivery and beyond, right?

143

Randall Stevens: I know,

uh, Carl, uh, when I was.

144

You know, looking back at the conversation

we had a year ago, one of the examples

145

you gave was, um, you know, we were

talking about, you know, synchronous

146

versus asynchronous communication and

how, you know, chat and email and all

147

these different forms of communication.

148

I, I agree with you.

149

I think, I think every time, uh, every

time somebody thinks something's gonna

150

come along and replace the other, it just

means we've just added another source

151

or a form of, of, of communication.

152

And you can't, it's hard to stop.

153

But, uh, my, my question would be, you

know, when, uh, I think in that last,

154

uh, conversation, you, you gave an

example where something had gone into

155

some litigation and they had looked

back and said somebody had given like

156

a thumbs up to, uh, to a message,

which was like a tacit approval, you

157

know, um, s so do you think, uh, are,

are you all experimenting or do you

158

think that AI is gonna be able to help?

159

Um, you know, 'cause you can

get a lot of data points.

160

Obviously you can get a lot of, uh,

information that on its own looks.

161

It is not, it is not in

in the right context.

162

Uh, and, and it seems like the LLMs,

that's one of the things that they're

163

really good at, is if you can get a

good enough context window, it's like,

164

okay, I can kind of boil this down.

165

Or, uh, so I guess my question would be,

are you all thinking about using those

166

kinds of technologies proactively so

that if somebody had, you know, let's

167

say somebody gave a thumbs up, should the

system say, do, do you mean that that's

168

approval or you like the, you know, what,

what does that, what does that mean?

169

So you could think

170

Evan Troxel: Or it says,

you shouldn't do that.

171

Don't do that.

172

Randall Stevens: right?

173

Or just like, just to

be clear, is this right?

174

Because you can imagine you'd want,

if it's something for approval and

175

then ultimately, 'cause we've been

doing some experiments with it too.

176

I think a really good use case of, of

the LLMs is to take, you know, kind

177

of sparse, uh, detailed information

and, and roll it up into things

178

that, you know, have meaning to

different people at different times.

179

So kinda what, what are you

all seeing on that front?

180

Carl Veillette: Yeah, so I think I will

just take a few, uh, a few use cases

181

of ai and we just rolled out our first

AI feature in, you know, last year.

182

So, um, I think AI can play

a role in multiple things.

183

It can play a role in collecting

information, which is exactly what

184

we did with our first implementation.

185

So I can talk a little bit about that.

186

I think it can also help, you know,

you're, you're thinking about,

187

uh, information at scale, right?

188

And you know, in my mind.

189

It's, there's no question about, about

the fact we're, we're living in a data

190

intensive, uh, world these days, right?

191

And I look at these, the size of projects

archive five years ago and what the

192

size of projects archive looks today,

the tools on the market have made it

193

so easy to create more and more data.

194

Uh, it, it's almost like we've reached

our human limit of how much information

195

that we can consume, crunch, analyze,

you know, whatever on the project, right?

196

So I think you would, you know,

you would think more tools equal

197

more, like more efficient workflows.

198

I have doubts about that actually.

199

When you look at the index of productivity

and construction, it's been flat right?

200

At past.

201

So I, I feel like the, you know,

those tools are making certain

202

activities more efficient, but it's

also like contributing to the over

203

being of information overwhelm, right?

204

So I, I think crunching and

analyzing information as.

205

At scale is only gonna be possible

through the power of LLMs.

206

'cause you would probably need like

10 people to look at the whole project

207

information, try to figure out the project

summary, where are things at, right?

208

But now with LLM, you can, you can

build like a search engine, you know,

209

on top of your, your data layer on

the project, and then you can start

210

to interact with that l and m and it's

gonna be almost instantly giving you

211

answers for millions of data point.

212

Right?

213

So I, I'm, I'm really fascinated

about like, you know, the ability

214

to, to summarize information, right?

215

To make informed decisions.

216

Uh, you know, search doesn't really

sound sexy when you say it fast, right?

217

But, uh, it, it, it is fascinating, like,

you know, the, the dimensions of like what

218

can search provide on the project, right?

219

So, um, you know, talking, you thinking

about like informed decisions, right?

220

So, um, think about like, um, uh,

code compliance for instance, right?

221

So like these days, uh, veterans

are retiring, people are

222

struggling with knowledge, right?

223

The corporate memory that they've,

you know, evolved and developed

224

over the last couple of years.

225

And you've got new, like you,

you've got those folks retiring.

226

You've got new folk coming out of school.

227

They're not experts in building goals.

228

You know, that knowledge that was, has

evolved, you know, from best practices

229

and knowhow throughout the years.

230

It, it's hard to transfer over, right?

231

And people are struggling, but what,

what is the role of AI in there?

232

AI can actually help onboard new

staff members by transferring

233

that knowledge, right?

234

So you think like, you know, AI these

days has trained itself on, on national

235

billing codes, uh, city regulations

and all those sort of things, right?

236

So what if, what if AI was brought in, in

the context of, you know, you're, you're

237

reviewing documents before they issued out

of construction and it's raising flags.

238

Oh, you forgot about this.

239

You forget about that, right?

240

Uh, you're about to issue something

that, you know that is not

241

compliant with the code right there.

242

There's miss a, a missing

fire device here, right?

243

That it doesn't meet the, the,

the fire protection, uh, uh,

244

part of the building code, right?

245

So think of it as, as a safety net, right?

246

But also to help, uh, potentially

train new staff members that, you

247

know, maybe there's an omission

on the, on the design, right?

248

And they, they didn't see it, right?

249

And now you're exposing yourself

for litigation down the road.

250

So I, I think those are, are

practical examples of ai.

251

Um, and I, I think like we're talking a

lot of AI about like the ability for AI

252

to crunch and, you know, find information,

search, you know, the chat GPT out there,

253

you know, Irv, we talks about that.

254

But AI can actually also play

a role in collecting data.

255

And that's exactly what

we did, uh, last year.

256

So we, we released an AI system that

actually help collect, uh, emails.

257

Uh, from your inbox.

258

So it's kind of like a smart scanning

system that stands through your

259

inbox, collect information, associate

them with the project automatically

260

so that you've got a decision No,

where every project record exists.

261

So again, when litigation kicks in

at the end of project, you've got all

262

evidences, you can fight back and,

um, be efficient during the course

263

of the project when searching for

information and, and, um, you know,

264

uh, documents shared and and so on.

265

Right?

266

Randall Stevens: So that's, so

that primary use case is, is a, uh.

267

A historical kind of use case

to crawl back through it.

268

Are you all thinking about though, how,

how you can use that technology to be

269

proactive so that there's not, well, I

guess in your example of, of training

270

and coaching along the way, that's

a good example of, uh, trying to be

271

proactive instead of reactive to, uh.

272

Carl Veillette: Yeah, no, definitely.

273

And I think we've got a

unique position because we're,

274

we're sitting on top of 20.3

275

million projects

delivered with our system.

276

So that's a lot of data, right?

277

And so the, the next step of our

journey is gonna be a Gentech

278

ai, and then we're setting the

foundation for that Agen Z system.

279

And so those agents would be

performing various different things.

280

Like I was talking about code

compliance, I was, um, you know, we've

281

got some other examples with W2 NRFI

responses, things like that, right?

282

Um, but, uh, we also see like

a, some, some somewhat of

283

an, a knowledge agent, right?

284

That would just basically, uh, pop

up, you know, and then tell you, okay.

285

Um, you're, you're designing a project

right now where you're, uh, you have a,

286

a green roof part of the design, right?

287

Uh, well, your firm has worked on, on,

on the green roof before, but it wasn't

288

called reroof on this other project.

289

It's a sedum roof.

290

Right?

291

But it's, it's the same, it's related,

it's semantic association, right?

292

So it would start to bring other

examples of, of, of that design and

293

bring additional documents inside, maybe

like that, that green roof, uh, ended

294

up, uh, having a bunch of like, change

orders or issues or additional RFIs that

295

came in because those information were

missing from the construction documents.

296

Right?

297

Um, so it would bring those, um, lesson

learned from other projects that we

298

don't repeat the same mistakes, right?

299

Um, so that's kind of like the, the

idea of where we're going with this.

300

Um, and, uh, you know, we're, as I,

as I mentioned, we're, we're sending

301

a lot, a lot, a lot of data, but

also it's the, it's the relationship

302

between, you know, the, the data

itself and what happened, right?

303

So we kind of need that golden thread to

be able to provide like those insights.

304

So, uh, you know, the, you run into

issues after construction, right?

305

So, um, if you don't have those.

306

You know, data point and those

relationship between the items, right?

307

Between that, um, that RFI or that

contract change item, you know,

308

that's related to the, the design.

309

It, it, it's hard to pre, it's

hard to predict, it's hard to

310

provide those insights, right.

311

But, uh, we do so.

312

Randall Stevens: Carl, we were talking,

uh, uh, in a podcast that we just

313

did recently with the guys from, uh,

twin company called Twin Master, and

314

they've been working on digital twin.

315

How to kind of, you know, kind of

broadly, how do you contextually

316

tie all this information together?

317

And one of the con one of the parts

of that conversation were, uh, around

318

like code and, um, uh, you know,

I just wanna get your thoughts.

319

It seems like that the, the, the

safe approach to that because nobody

320

wants to claim that they've, that

they're, you know, giving you the

321

right information is to, is to let the,

the customer bring that information.

322

To the system and then, and then let

the system use that as a source of

323

information to give information back,

as opposed to you saying, we've, you

324

know, we've got that information.

325

How, how do y'all think about that?

326

'cause you know, as you know, code

compliance can be very specific to

327

Carl Veillette: Yeah.

328

Randall Stevens: jurisdictions.

329

Carl Veillette: Yeah, actually, I,

I had a quick chat with, uh, the

330

guys from UP Codes recently, and,

uh, they have a great platform.

331

It's, uh, you know, building

a very extensive, uh, building

332

code, uh, library, if I can say.

333

Right.

334

And they've got some AI to make

some searches in there, right.

335

So, um, yeah, we, we had an interesting,

very interesting conversation, you

336

know, about, um, some, some, some, um,

uh, some litigation that went on around

337

the, you know, the building codes and,

you know, should you be able to, uh,

338

you know, uh, train and build AI and

ownership of like the, the law, right?

339

And those sort of things.

340

But, um, um, you know, I, I think they

would make a, a good partner, like

341

for the things we have in the roadmap.

342

So, uh, you know, being able to tap

into their, the library of codes is not

343

something, I mean, trying to maintain a,

a code library is not something that is

344

new format's mission, that's for sure.

345

Um, but um, uh, you know, like

kind of, uh, leveraging that.

346

Um, to, to be able to apply it to

specific construction documents, you

347

know, visual artifacts and things

like that is something that we're

348

definitely, definitely exploring.

349

Right.

350

So, um, uh,

351

Randall Stevens: I haven't

352

Carl Veillette: playing different roles.

353

Randall Stevens: yeah, I haven't kept up

with, uh, with what their strategy was.

354

But are they, uh, you

know, stop top of my head.

355

It would seem like if you could end up.

356

You basically befriending the

jurisdictions because they're

357

the ones doing the code review.

358

So if you could go to the city and say,

actually we've got a system that will

359

not only take your requirements, but

check the incoming stuff coming in, and

360

that way you would at least have, you

361

Carl Veillette: Yeah, no, I think,

I, I would envision, you know, uh,

362

we could geek out on this, right?

363

But I would envision any integration

between Newforma 'cause we're

364

managing the workflow, right.

365

For issuing construction documents

in our system through that control.

366

Right.

367

I would imagine like their code tying

into like, we've got the project

368

address location and such, and then.

369

You know, I, I could easily envision

like there, you know, they would be

370

suggesting the, the applicable codes

and then would be, you know, kind

371

of building some AI around that to,

uh, uh, kind of analyze the, uh, you

372

know, the, the construction documents.

373

So, um, I mean, we, we own the process

already and they own the library, so it,

374

you know, it would be a great partnership.

375

Evan Troxel: Oh, I just want

376

Carl Veillette: my request of codes.

377

Evan Troxel: i, I want

to give a shout out.

378

I, I love, I love what O Up codes is

doing, and I also want to give a shout out

379

to another platform that's doing exactly

what you're talking about, because I don't

380

think UP Codes is doing that, but it's

called Archie Star and they're working

381

with specific jurisdictions because

it's very difficult to roll that out.

382

I've talked to other people who say

there's, in the US alone, there's

383

30,000 different jurisdictions that

have their own version of code adoption,

384

and so they have different years of

code adoption, different, you know,

385

different layers of code adoption.

386

And so it's super

difficult problem to solve.

387

Um, but Archie Star was doing that.

388

For that exact reason that

you just mentioned Carl.

389

It's like, and, and Randall too.

390

Like it's, it's, you're working with

the jurisdiction to say, we wanna

391

take the load off of you and we want

to help the design professionals

392

make this whole process go faster.

393

And I think you're right, Carl, like

pairing that with the project data,

394

like you're talking about, and, and

not just current project data, but

395

project data throughout the firm.

396

Right.

397

And, and so I guess my question is, is

with, with a, with a tool like Newforma

398

and, and kind of the, the depth to

which it goes, I think the hard part

399

for bigger firms has always been user

behavior to get, and, and I know you,

400

you released this AI tool to kind of mine

the information autonomously because why?

401

Because users suck at this.

402

Like users are, are notoriously bad

at archiving the data and archiving it

403

properly for their future selves because.

404

Every project is a new startup, right?

405

It's a new team, it's new constraints

for the site, it's a new project brief.

406

It's all of these things.

407

And so a lot of times we

just file bankruptcy at the

408

end of every project, right?

409

Not literally financially, but,

but literally as far as the

410

team goes and knowledge goes.

411

Um, and, and then we start fresh.

412

And so I, I'm curious how you guys

think about that and how it seems

413

like you're thinking, well, AI can

help with this, but I mean, you've,

414

you've seen this before, right?

415

That this kind of weak chain link

that is in the system, which is

416

the actual user in their behavior.

417

Carl Veillette: Yeah, so I, you know,

I, I, I gave the, the example with

418

the, you know, the green roof earlier.

419

Right.

420

Which is, you know, okay.

421

Trying to, to gather information.

422

I'm gonna give you another example of

like some of the things we, we have, um.

423

In the kitchen right now.

424

So like I was talking about search, right?

425

But when you've got like a new intern

coming in, in the, in the firm and

426

they are trying to figure out what

projects have been delivered by the

427

organization before their time, right?

428

It's not an easy thing.

429

You think about project

proposal right there.

430

There's an, there is a,

there is an RFP, right?

431

And then this in turn is gonna have to

kind of like do a little bit of research

432

exercise here, try to figure out how many

hospital projects they delivered before.

433

Where are the files, right?

434

That might be in a, you

know, A CRM somewhere.

435

They might be on the SharePoint.

436

They might exist in

multiple locations, right?

437

So one of the things that our

system's gonna be able to do,

438

because we're connecting with

those various data source, right?

439

We're tapping into SharePoint,

we're tapping in local fast servers.

440

We've got connectors for

all of those things, right?

441

You're gonna be able to just go in there

and ask, give me the list of the proposals

442

no matter where they live, right?

443

Um, for the, the hospital

projects that are above a

444

billion dollar in value, right?

445

And it will just give you other

proposals, okay, now help me

446

write a proposal for this.

447

Here's the template, right?

448

And then, you know, bring the

resume, bring the, you know, so,

449

so those things exist, right?

450

They're just like, it's sprawl, right?

451

It, it's, it's living in different silos.

452

It's all over the place.

453

There is no process for it.

454

Right?

455

Um, so it, it can really help, you know,

not just a, not just the training, but the

456

onboarding of those new staff member and

tap into that collective knowledge that's

457

been created, um, by the organization

for the, the last decade or so.

458

Right?

459

Um, so I, I just, I just think about

like how much time would be saved in

460

that context as they write that proposal.

461

Right?

462

Um, so there, there is, um.

463

Yeah, there's a startup in, uh, Toronto.

464

It's called, uh, work Orb.

465

That's what these guys are doing.

466

It's pretty cool.

467

And they have connection with CRMs.

468

And so, you know, I can see AI playing

a big role in both the onboarding, the

469

training, the quality control, adding

additional security nets, you know,

470

um, around more junior staff members.

471

Um, so that, that's a, that's a

very exciting time because it's

472

multidimensional, if I can say.

473

Right.

474

So, um, uh, I think, um, I think

there's gonna be like, the more we

475

think about it, the more we're gonna

find, like, uh, you use cases for

476

almost every activity of the project.

477

Um, you know, just, uh, just

think about legal, right?

478

I was talking about the last

phase of construction project

479

being litigation, right?

480

Um, it's imagine you're, imagine

you're reviewing documents, right?

481

And then something pops on the

side of the screen says, uh,

482

you've run into litigation.

483

Uh, about that, you know,

on previous project.

484

Um, so pay attention to this.

485

Like, here's like, you know, how

much it, you know, this, uh, this

486

created in terms of costs, you know,

on the project level, and it was aate

487

whether this or this was good, right?

488

Um, and, uh, you know, the, the

firm ended up, uh, losing the, the

489

case, you know, for this, right?

490

So, uh, you, you, you could also like,

you know, bring those type of insights

491

in the context of the project activities.

492

So that's, uh, that's very exciting.

493

Randall Stevens: Carl when you

494

Evan Troxel: just want to just real quick.

495

I just wanna say that would be

incredible because again, it's not

496

even the same team that was on that

other project where litigation happened

497

and the current team doesn't know.

498

Right.

499

And so that would be

incredible to be able to source

500

Carl Veillette: it's almost like

we're not building planes here, right?

501

It's almost pro every project is

different and then the project starts.

502

You've got a, a set of different teams.

503

They have to learn how

to dance together, right?

504

If I can say, right.

505

So.

506

It's almost always like a new kind of

learning process and how we're gonna

507

collaborate, which tool we're gonna use.

508

Um, so it, it's, um, it's a, it's

a very dynamic environment and that

509

creates a lot of like uncertainty on,

you know, the, the predictability of

510

the, of the project outcomes, right.

511

The schedule, the, you know, like the

deliverables, quality and such, right?

512

So, uh, I think it's, uh, you know, if

we can, you know, if we can alleviate

513

some of those through the use of ai.

514

Uh, and just thinking about like the

data intensive era that we're living

515

in right now, uh, if, you know, AI can

certainly help us stay on top of it.

516

Randall Stevens: a little bit of, I guess

maybe a little technical question just to

517

think, to hear how you think about this.

518

So, when, when you talk about feeding

this information into the data, you know.

519

Bringing the context windows and, and

all this information in do, can, can

520

you actually just dump everything in?

521

Or do you need to think about a kind

of specific additive process for the

522

information that you know is va you know?

523

Correct.

524

Because there can be bad, you

know, incorrect information.

525

I was just thinking when you were

talking about training, you know, a

526

a, a young professional that's just

come in if, if they have access, you

527

know, if you're pointing to everything

that's been back there, not everything

528

necessarily is even correct or good.

529

How, how, how are you

all thinking about that?

530

And, uh, 'cause it is,

it's obviously easy.

531

Everybody wants to, everybody

wants to just throw everything.

532

I just, uh, I just, uh, had, uh,

somebody on our team say that we've

533

got a customer that's wanting to

point at 12 million files to, you

534

know, dump into, into our system.

535

And it's

536

Evan Troxel: Because

more is better, right?

537

Randall Stevens: well.

538

It usually is, and, and you know,

anybody, you know, and I'm guilty of

539

it too, it's like, well, if you're

a technologist, you like to, you

540

know, you also like to break things.

541

Like, okay, I'm gonna push it to its

limits and, and see where the edges are.

542

But, uh, but just kind of back to,

you know, this idea that you all do

543

have all those projects and all that

dense information, you know, you're

544

talking about millions and millions

of files over time inside, across,

545

you know, thousands of projects.

546

You know, do, do you have to take, uh,

if you're the CIO sitting in one of those

547

operations, do you have to begin to think

about plucking out, you know, specific

548

information to feed into the system?

549

Or do you think that the AI tools

that we're either have in front of

550

us or near term are gonna be able

to help us discern what was good and

551

bad information or help to do that?

552

What's, what's your thoughts there?

553

Carl Veillette: I mean, I, I, the way

I think about it is that, you know.

554

Garbage and garbage out, right?

555

So, I mean, uh, just dumping

a, a lot of unstructured data

556

to when AI is quite risky.

557

So it's like at the end it's just

think about a specific use case, right?

558

Like, I mean, let's say, um, you're

building AI that's gonna be, uh, helping

559

GCs on site, uh, find information

from construction drawings, right?

560

Okay.

561

Uh, well, construction drawings, there

maybe multiple revisions of it, right?

562

So if you don't have a good

structure, it's easy for AI to

563

get confused and give you the, the

answer from the previous revision.

564

And then all of a sudden you've got

an issue with, you know, the, the,

565

the concrete pour on site, right?

566

Uh, didn't put the right

rebar, you know, whatever.

567

Right?

568

So I do feel like the, the structure

of the information, uh, that

569

you're feeding ai, uh, with right?

570

Is, is very important.

571

Um, I almost think about the

project as a, there's like.

572

The, the way we're approaching

things is everything is times

573

stamped in our system, right?

574

So often, you know, where AI is gonna

get confused, it's, it's gonna take a

575

superseded set of drawings, or it's gonna

be taking conflicting information from

576

an email and it's gonna be superseding,

you know, with, you know, the answer from

577

the data's construction documents, right?

578

And things like that.

579

So everything being time timestamped, you

can actually create like a, a timeline on

580

the project, like a conceptual timeline.

581

So, uh, you know, this is how we've

structure the information so that

582

we ensure that, you know, competing

information, um, you know, is, is,

583

is not confusing the ai, right?

584

Um, so I, I think, you know,

having a good data structure behind

585

the scenes is very important.

586

Uh, you know, if you take, uh, if we,

if we, if we took our project archive

587

as is and we just dumped that to a

chat GPT, and we start to ask Chachi PD

588

questions, it's most likely it's gonna.

589

Provide you wrong answers.

590

It's gonna use a, a wrong document

or it's gonna use an email where

591

some information wasn't official,

but it was shared in there.

592

Um, so, so at the, at the end, I think

it's very important to, to look into

593

the, the backend system that's created

to manage the data in the first place, to

594

collect the information, to structure it.

595

Randall Stevens: I, I

would, I would agree.

596

But I also, I've been trying to train

myself because, you know, I, I've been

597

in, you know, doing technology now for

35 years and I'm fond now, you know,

598

the, the old saying that really good

software is indistinguishable for magic.

599

It should really be magical.

600

And we're watching, you know, I'm

using some of these tools and I'm like,

601

I, it, it's, it is really amazing.

602

Evan Troxel: magic than it's ever been.

603

Randall Stevens: Well, and it's magic.

604

And I've got a, uh, well we had, uh,

we had a friend of mine who's, who's

605

one of the world's, you know, renowned

machine learning and computer vision.

606

Guys who, you know, keyed

CCV this, this past year.

607

He told me, I don't, I don't, you

know, when he tells me he doesn't

608

understand it, I have a little solace

in like, 'cause usually, you know, if

609

it's technology, I wanna learn enough

about it to at least have a construct

610

to say, okay, I can kind of understand

what it's doing and how it works.

611

I think we're in the middle of, of this,

um, you know, of a new wave where we

612

can't even wrap our heads around it.

613

So what I was gonna

say about the, I agree.

614

You know, my, my tendency is to say

you want good structured data to do

615

this, but I also say, well, if I gave

that to a human, what, how would they

616

decide what was relevant or what wasn't?

617

And if a human can do it.

618

Why do we think that we can't

have an AI agent doing that too?

619

So, you know, it may not be today,

but maybe it's just around the

620

corner, then it's like, maybe it

can, 'cause I mean, I think we have

621

to ask ourselves if a human could do

it, why couldn't the AI engine do it?

622

'cause I think that's what wave that

we're in right now of this, you know,

623

fast rolling, uh, kind of updates and

changes to the way this technology works.

624

So anyway, I'll just

throw that out there like,

625

Carl Veillette: Yeah, no.

626

It's funny because like, you know,

I, I hear a lot these days, uh,

627

across all industries, right?

628

Like.

629

There, there is a, there is a sentiment

that AI is gonna disrupt, like large

630

enterprise softwares like Salesforce

or Oracle or things like that.

631

And, you know, you can see it,

you know, you're gonna, the stock

632

market and you know, the, these

stocks are crashing right now.

633

It's just, so I, I'm like, this is

really strange because AI is not gonna

634

disrupt enterprise software because

those software have data and, you know,

635

like AI is only as good as data that

it's, it is being trained on, right?

636

So I think where the value is, is data.

637

It's not the AI itself.

638

Um, now AI is a mechanism to crunch the

data to make something with it, right?

639

But it, you know, I guess like the

world we live in at your FOMO is

640

like, we have this data, right?

641

So like, you know, we're gonna be building

like things with AI on top of that, right?

642

But at, at the end, this is,

you know, this is, this is where

643

everything converge, right?

644

If, if you have data,

there's value, right?

645

So you can, you can do something with

it, but if you don't have data, then.

646

You're just gonna be like,

the flavor of the data.

647

It's like there's so

many of those AI tools.

648

Like at some point you may ask

cloud AI to build an app that

649

crunch, you know, documents for,

for you, for code compliance, right?

650

And we may not be that

far away from this, right?

651

So people are gonna be able to

create their own AI flavor, but if

652

you don't have the data, then you

know, you, you can't, you can't

653

train any, you cannot train on it.

654

So that's, that's where

the value is in my mind.

655

It's, you know, who owns the data, right?

656

Randall Stevens: So let's talk,

uh, let's talk text stack survey.

657

What, uh, I, you know, I was looking

back, uh, when we put that thing

658

together, we had, uh, you know,

there was over 400, you know, kind of

659

applications that we kind of knew about.

660

And, and literally it was go, let's

just do a, see how many people click

661

and say, I'm using that, I'm using that,

we're using that, we're using that.

662

And literally just to,

uh, take an inventory.

663

There was a little bit of nuance

to it in that for each one that

664

they were using, we set, we ask

on a scale of one to five, is it.

665

Deep penetration, you know, deep

in the workflow or, or just here

666

and maybe not being used very much.

667

And that was at least a, a, a, you

could read that in a lot of different

668

ways, but it actually, as we pulled

that data, you could do quite a bit of

669

kind of analysis across not only the

tool stack, but that kind of sentiment

670

for how deeply rooted that things are.

671

What I was gonna say though is we

broke those apps up, those two, 400

672

apps up into 20 different categories.

673

I think Newforma ended up in

the same category as we put our

674

Carl Veillette: That's really strange.

675

'cause when I, I think about

our respective platforms, we

676

don't do the same thing at

677

Randall Stevens: Yeah,

completely different.

678

And, and that's what, you know, I, I, my

my excuse is it was the first year we did

679

it and we were just trying to figure out

like, okay, how would we, uh, break this

680

up into some, you know, broad categories.

681

There's probably some nuance that we

need to add the next time we do it right,

682

which would be, so that was gonna be my

first question is what category, what.

683

What, what kind of broad area would

you call the, where Newforma lives

684

and what other applications kind of

live in that same, like is Autodesk

685

part of Autodesk platform would be

considered along those same workflows or,

686

Carl Veillette: yeah, so we play, we

play in a diff it's a, we, we got a, a,

687

a strange position because we're, um,

we're playing on different fronts, right?

688

But we're not doing the actual work.

689

So let me explain that a little bit.

690

So, in your survey, for instance, there

was a communication category, right?

691

We're not.

692

We're not Microsoft teams,

we're not Outlook, but we're

693

connecting with those, right?

694

So we're like, when it, when it comes

to email management and those sort of

695

things, we've got some, some of the best

in class features for that, but we're not

696

Outlook, we're not distributing emails.

697

Randall Stevens: not the core application.

698

You're the connective

tissue between those,

699

Carl Veillette: Exactly.

700

And that's, you know, we're

talking about, we're talking

701

about the golden thread, right?

702

And I think that's kind of the role

we're, we're playing, we're, we call that

703

project information management, right?

704

And we've been calling it this way for 20

years and it's almost like we're category

705

makers in this space because there

isn't, you know, other solutions that,

706

that, that do that to the extent we do.

707

So we connect with ERP system.

708

We're not an ERP system, but

we're gonna be enabling, just like

709

knowledge architecture does, right?

710

Feeding from like ERP systems to

create projects so that people

711

don't have to do it twice, right?

712

So the information

flows, it's being reused.

713

We recycle information.

714

Same

715

Randall Stevens: we do the,

716

Carl Veillette: administration,

717

Randall Stevens: we do the same.

718

We have the same problem.

719

We, we call our, you know, by default we

call it a content management platform,

720

but usually most people think of that.

721

They think that we're not gonna be the.

722

Take your that information, but really

what we do is connect in very much the

723

same way you all do to where those pieces

of information are and basically fuse it

724

together into one, you know, kind of view.

725

Uh, is is where our value add is.

726

So what do we call, what do we

what, what should we call that next?

727

Next time we do it, is

728

Carl Veillette: yeah.

729

I, I, I, I, I

730

Randall Stevens: bridge, bridges,

connective tissue, bridges, or,

731

Carl Veillette: project

information management.

732

Randall Stevens: yep.

733

Carl Veillette: there, there's

a few players out there that,

734

you know, use that umbrella, uh,

term for, for similar things.

735

Right.

736

Um, you know, I would, I would argue

that we, we also, uh, do a lot of like

737

construction administration, right?

738

We've got a lot features around that.

739

So.

740

Uh, that might be an overlap with

Autodesk, for instance, right?

741

We're certainly not competing

with Pole Core 'cause they

742

do construction management.

743

So they're from the general

contractor side of things.

744

We take it from the designer,

uh, perspective, right?

745

Uh, so we actually have a good partnership

and connectors, you know, with Pole core.

746

Um, so, uh, people love it in general.

747

So we were playing on that front too.

748

Um, think about bim, right?

749

We're not the clash detection software.

750

We're not, you know, um, we're not

an authoring system either, right?

751

But we've got some pretty unique,

you know, uh, issue tracking

752

features and functionality.

753

So I, I guess we do play in the

BIM coordination kind of space.

754

So you look, there's a lot

of softwares in that space.

755

Um, so again, with information

and management in mind,

756

we're not detecting clashes.

757

We're the bridge between, you

know, those in celebrity software

758

to the ordering system, right?

759

So we relay the information,

we make it accountable.

760

We create these three, the

audit trail, um, we make the,

761

the workflow more efficient.

762

But, um, at the end, we're not a differing

system either, but we're, we're also

763

living there as add-ons and, and add-ins.

764

Right?

765

So we're, we're kind of

connecting the dots, right.

766

So that that's the role we wanna play.

767

Evan Troxel: think it's interesting that

you're explaining all these different

768

places that you play in, and I bet most

of your users know you for one or two of

769

those things, not all of those things.

770

And so like, I think this kind of

comes back to the idea of what a

771

brand is known as or known for is

really up to the customers and not

772

the company that makes it itself.

773

Because if you asked me what new formal

was and you asked Randall what new formal

774

was, we might have two different answers.

775

And somehow you might want

to try to reconcile all that.

776

I don't know, maybe you don't,

but, but like when you ask that

777

question, Randall, I kind of think

you want to go to the users and

778

say, what would you put Newforma in?

779

And you're gonna get a

lot of answers, right?

780

And then it's the job of

try to reconcile it, but.

781

It's like when you say, you know,

veils content management, you know,

782

and, and what would be interesting to

see if people think of it as something

783

different than just content management.

784

Randall Stevens: Yeah.

785

As we were, you know, and really,

you know, we were, uh, we're not

786

professional surveyors and we

were kind of doing this on a,

787

Evan Troxel: You're not dodge

788

Randall Stevens: last fall.

789

Yeah.

790

Yeah.

791

So there wasn't a ton

of resources to do it.

792

So, no, I, you know, we,

we were, we did our best.

793

Uh, but, uh, in hindsight, you know.

794

By trying to split this stuff up.

795

We, we were trying to put each application

in one of the 20 broad categories, and

796

in reality, like you're saying, Carl,

it's like, oh shit, you, you could've

797

been a little bit over here, and a

little bit over there, and a little bit

798

over there, which then if somebody had

taken the surveys in that section, it

799

might've been like, you know, they were

probably just responding to what they did.

800

We did have a, we did have the ability

to write in, you know, so if we were

801

a glaring mistake, they would write in

like, how come you didn't have X, Y, Z?

802

But, uh, but it also could have just been

like, Hey, we, we didn't just, we're just

803

kinda responding to what you put in front

of us as opposed to really thinking about

804

if I ask you without any of that in front

of you, you might've said, oh yeah, we use

805

Newforma and Procore for doing that work,

or, you know, those parts of the flow.

806

So I think that's gonna be a challenge

for us to think about how to do it.

807

But, you know, like anything, it's

almost, you have to put a stake

808

in the ground and say, I, I wanna

create these kind of buckets.

809

And it doesn't mean that they're perfect,

but at least gives a conversation starter.

810

Carl Veillette: it, it's like

what's a collaboration software?

811

Is that a characteristic of, of, of a

workflow or is that an electoral category?

812

Right.

813

I mean it's, you can ask yourself

that question 'cause like almost

814

everybody these days, like

have collaboration features.

815

And I mean, as soon as you've got

multi users collaborating together,

816

is that called collaboration?

817

Like what's the true definition

818

Randall Stevens: You can

share a link and avail.

819

So is that

820

Carl Veillette: yeah.

821

Is that collaboration?

822

Randall Stevens: Yeah.

823

Uh, just, uh, you know, part of, there

was actually two categories that kind

824

of related specifically to what we

do, and I think to where what you do.

825

And, and that was, you know, basically

where are the, where are the primary,

826

I'll just call it data storage location.

827

So what does your file system look like?

828

Are you using SharePoint?

829

Are you on a, you know, a c, c, or,

you know, there's places where the,

830

there's literally file storage.

831

And then there was another category

that was more like content management

832

or that, I'll say a layer above it,

which is, we use this, but some people

833

would, would consider SharePoint that.

834

Right?

835

They think of it as like a website.

836

You know, back to what you were saying,

Evan, I think it's like to the user,

837

that's not a storage location, that's

like a, a website, you know, because of

838

the way the interface to SharePoint is.

839

But yeah, like, you know, I

always took Remind people is

840

just one drive back there.

841

It's like, it's just, it's just files

being stored somewhere and then you've

842

got some skin on the front of it.

843

But what we did find from, from, you know,

if you take both of those, uh, sections

844

of the survey across, you know that across

the, you know, a hundred plus people that

845

responded to it, they averaged about a

dozen of those locations in use, which

846

part of what we were wanting to show was.

847

There is data sprawl.

848

You know, we're, we're not only

producing more information,

849

it's living in more places.

850

And now the end user has to know, is

it over there or is it over there?

851

Is it over there?

852

And, you know, I think that that's

becoming a bigger and bigger part,

853

which is what part of the problem

that we're trying to tackle.

854

And I think you all, you know,

operate in the same space, which

855

is like, look, this, this data is

living in a lot of systems of record.

856

How do we, how do we fuse

it all back together?

857

Carl Veillette: And I was looking

at avail, and it looks like you

858

guys are doing it pretty much like

the, the same, uh, thing that we do.

859

So we, we don't really wanna

host like all the data, right?

860

So we're kind of like connecting with

those various different data sources.

861

So we connect SharePoint, a, c,

c file, like local file server.

862

Um, we, we have Ignite, you know, and,

and those, you know, other players like

863

Panzura and ADI and, and such, right?

864

So we, we're not a storage platform.

865

We're not an EDMS, but we're

providing a view into those things.

866

So, uh, as information is being shared

outside the organization, there are

867

proper retention policies for files.

868

You can populate document control

so that you can meet standards

869

like ISO 19 60, 50, right?

870

And things like that.

871

So kind of like ensuring the

compliance and the government's

872

layer on, on top of those various

different locations that exist.

873

Um, so like we're kind of trying

to place Switzerland in a way

874

that we're neutral, right?

875

And we're, we're gonna be connecting with

the various data source using the project.

876

And, you know, you think

about infrastructure projects,

877

people are gonna have files on

ProjectWise, well guess what?

878

We connect with ProjectWise.

879

Um, they're gonna have some

maybe files on that same project.

880

They're gonna be Revit

files who sit on acc, right?

881

Um, so when you're using all those

systems independently and you're

882

trying to manage like issuance and

things like that and get proper

883

audit trail, it's quite hard, right?

884

Um.

885

Randall Stevens: I know you, uh, you're

a relative newcomer to Newforma, but, uh,

886

how, how long do you think that, that,

that term governance has been in use?

887

In just in general?

888

'cause I would think new form.

889

If it was gonna be talked about, you

all probably would've been one of

890

the early ones to to, to at least

conceptually talk about what it means

891

to govern kind of information across

any, any idea of when that came into

892

the vernacular in this industry.

893

Carl Veillette: Yeah.

894

No, I, I, I think, you know, uh, we,

as we saw the industry evolve, and

895

I think, you know, AI and cloud has

certainly like, um, I would say woken

896

up a lot of people on governance, right?

897

Uh, and, uh, I think the

proliferation of tools has made

898

it, uh, a thing in the industry.

899

I, I'd say, you know, for the

last five years I've seen.

900

It and CIOs and organizations trying to

get a hold of their data and understand

901

how the data flows in the organization.

902

And, um, I think it has

become a, a top concern.

903

And I guess like the industry

report says so, uh, as well, right?

904

That that's becoming

an increasing problem.

905

Um, and especially because

data is gold, right?

906

So, and in these, uh, in that

AI era that we're living in now.

907

So, uh, I think a lot of people are

getting, um, other concern about where

908

the, the data is and concern about their

ability to tap into their own data and,

909

and how they're connecting it and how

they're consuming it right down the road.

910

So, um, organizations are, are

looking at strategies for that,

911

uh, in our timeframe, right?

912

I think, you know, you guys do it, you

know, you may look at the, you know,

913

co the content like li libraries,

like details, things like that, right?

914

So it's, it's really cool, uh,

to be able to reuse information

915

like that coming from various data

source and, and leverage that for.

916

Uh, for, for making like ongoing

projects more efficient, right.

917

Uh, that's just, you know, one

example of, of, um, uh, what

918

technology can do these days.

919

Um, um, but yeah, I'd say, you know,

governance in general, um, in your format.

920

I think we, we, we see also that, that

there's a lot of regulations, you think

921

about the Building Safety Act in the UK

and those sort of things which require,

922

um, uh, decision logs and things like

that to be, uh, documented, right?

923

So, uh, regulations are also

pushing organizations to get better

924

governance in place, uh, not just

from a series standpoint, right?

925

But from, um, uh, project records, right?

926

Uh, you think about contracts, right?

927

You've got retention policies for data,

certain contracts, uh, you know, it's been

928

the case for the last two decades, right?

929

Uh, more actually.

930

Um, so how do you make, like, how

as you grow your tech stack, can you

931

really, truly get, turn into a contract

where you've got project retention

932

policies and after X amount of years,

like you gotta delete the data.

933

Like, it's, it's not

934

Randall Stevens: Yeah, that's tough

935

Carl Veillette: It, it's tough, right?

936

Randall Stevens: Yeah, I've had

those conversations because a lot

937

of, you know, you take something like

Miro, which is, you know, being used

938

quite a bit, but it's like it is.

939

Cloud first.

940

It, you know, you're in,

you're in that database.

941

So what does it mean to archive that?

942

How do, how do you archive it?

943

What if that, you know, you can

keep your account and they may keep

944

your boards around, but what if they

aren't around in five or 10 years?

945

How, what does it mean to

archive that kind of information?

946

And I think it's a,

947

Carl Veillette: Yeah.

948

I.

949

Randall Stevens: think

anybody's figured it out

950

Carl Veillette: Yeah, I'm just

thinking like it is fun facts, right?

951

Uh, like we used to have backups for

seven years worth of data, right?

952

And recently we had, we had to shorten the

period because seven years is too much.

953

I mean, if, uh, you know, if a, if a

customer has retention policies in place

954

for project, maybe the, their policy

internally is to get rid of the data

955

after the retention policy is set, right?

956

Maybe that's because they're worried

about the growing storage costs.

957

Uh, maybe they just don't wanna

have the information when litigation

958

takes in beyond that date.

959

And, and I've seen examples of that.

960

Evan Troxel: Sure.

961

Carl Veillette: Yeah.

962

So, so some want, want, want

to keep it as a, you know, as a

963

backup to be able to fight back.

964

But some they just don't wanna

have the data in their hands.

965

So, uh, yeah.

966

So we ended up like, uh,

realigning our, our backup.

967

Um, uh, linked, you know, for

retention policies, purpose, right?

968

So, you know, it's easy with

I, Microsoft and AWS these

969

days to have like, just change.

970

Okay?

971

Now I've got 10 years

worth of retention there.

972

And it's gonna, because you have it,

it means you have to provide it, right?

973

So if you know it's kicks in

again, like the court's gonna

974

ask for it, well guess what?

975

It's in there.

976

Randall Stevens: yeah.

977

There's a litigation site.

978

It reminds me though, uh, of a, uh,

years ago I had a friend that worked.

979

It was a, he was an engineer

and, uh, worked at a, at a

980

pretty good sized company, and

he told me the story about that.

981

Uh, you know, people started

just in meetings saying, you

982

know, uh, well, you said this.

983

You know, six months ago, so they started

to have this policy that said if it

984

was more, if it was said more than 30

days ago, it's like it, it's as if it

985

didn't happen because things change.

986

So it's like you can't point back

to something I said a year ago.

987

The context may have changed or, you know,

988

Evan Troxel: I mean, this gets back to

your question earlier, Randall, like,

989

like with with, if somebody throws 12

million pieces of data into a, into

990

a thing that they want to cut cross

sections through and get insights

991

from, it's the same question, right?

992

Like a lot of that is old information.

993

And so like, what I like to think of

that as is like what the concept is.

994

What I think should happen is information

should have an expiration date.

995

Just like laws should

have an expiration date.

996

Right?

997

Or at least let's revisit them.

998

But no, it's just additive, right?

999

And, and the old stuff's in

there and it's just data.

00:51:31,071 --> 00:51:35,211

And so to qualify that I think is

kind of the, it's a very difficult

00:51:35,211 --> 00:51:40,131

thing to do, but, but same goes for,

what did you say in the meeting?

00:51:40,461 --> 00:51:43,371

30 days plus ago doesn't apply anymore.

00:51:43,671 --> 00:51:47,691

Well, chances are that old data,

you know, it might have a different

00:51:47,691 --> 00:51:51,411

usefulness, but it may not answer

the question that's being asked.

00:51:51,411 --> 00:51:56,661

And so that, that whole context is so

incredibly difficult, like challenging

00:51:56,661 --> 00:51:59,361

to address, but also extremely important.

00:51:59,556 --> 00:52:04,716

Randall Stevens: I will have to, uh,

I've been working on a, um, a concept.

00:52:04,716 --> 00:52:05,706

I'm writing a white paper.

00:52:05,706 --> 00:52:09,636

I, I'd love to share that with you,

Carl, uh, and, and get your feedback.

00:52:09,636 --> 00:52:12,336

But I've been talking about

it as capital resources.

00:52:12,366 --> 00:52:15,071

You know, from a, from a business

standpoint, it's like you want

00:52:15,071 --> 00:52:18,366

to have, you wanna have your

information, you want it to have value.

00:52:18,366 --> 00:52:21,816

So thinking about it in terms

of like capital resources, but

00:52:22,206 --> 00:52:23,376

part of the thinking is that.

00:52:24,786 --> 00:52:27,396

As you create these pieces of

information in whatever form they end

00:52:27,396 --> 00:52:33,006

up living on, we're missing the right

contextual wrappers of data around that.

00:52:33,666 --> 00:52:37,416

Maybe it's an expiration date,

maybe there's certain parts of

00:52:37,416 --> 00:52:41,136

this that can have value over time,

but other parts not over time.

00:52:41,286 --> 00:52:45,636

And I think, you know, I think what

our challenge is gonna be as, as the

00:52:45,636 --> 00:52:49,536

people developing these technologies is

those systems have never been in place.

00:52:49,536 --> 00:52:52,926

You know, back to the tech stack

survey, you got a hundred to 150

00:52:52,926 --> 00:52:54,906

applications, and you know what?

00:52:54,906 --> 00:52:57,396

They're pumping out information every day.

00:52:57,516 --> 00:53:02,496

And it's like, but we're missing,

you know, I, I've been saying for

00:53:02,496 --> 00:53:07,686

years we've been missing, um, the

idea of a, of a content router.

00:53:08,076 --> 00:53:12,336

Like we think we, we have technology,

our networks have routers, and there's

00:53:12,336 --> 00:53:14,046

rules about that information as it goes.

00:53:14,076 --> 00:53:15,756

Those bits go across and

where they should go.

00:53:15,756 --> 00:53:21,426

But we've never had, we've never had this

concept of, as we produce information from

00:53:21,426 --> 00:53:23,586

all these systems, there's nothing that.

00:53:24,096 --> 00:53:27,456

One says where it should go,

but I think you can extend that.

00:53:27,546 --> 00:53:31,536

Not only where do I wanna store it,

but what is the contextual wrapper of

00:53:31,536 --> 00:53:37,176

metadata around that that lets me now

know what to do with that in the future.

00:53:37,176 --> 00:53:41,046

It's all, you know, it's buried

inside the files, which complicates.

00:53:41,046 --> 00:53:42,696

It's like, so anyway, I

00:53:42,801 --> 00:53:45,231

Evan Troxel: Especially when it's

in a proprietary file format.

00:53:45,231 --> 00:53:45,531

Right.

00:53:45,531 --> 00:53:48,711

So you may not even have

that application any longer.

00:53:49,176 --> 00:53:54,111

I, I just, just an anecdote from a, a,

a digital practice leader at a firm.

00:53:54,141 --> 00:53:57,856

I mean, they, they talk about this idea

of, you know, data sprawl that you've

00:53:57,856 --> 00:53:59,931

brought up a few times on the show and.

00:54:01,291 --> 00:54:04,866

The, the practice is

like, we can't control it.

00:54:04,866 --> 00:54:08,826

There's too many people, everybody's

got their way of doing it.

00:54:08,856 --> 00:54:12,186

They all have their own

preferences, which brings so much

00:54:12,606 --> 00:54:14,856

chaos to this process, right?

00:54:15,366 --> 00:54:21,546

And so they just said, ultimately we've

decided that it is the project managers.

00:54:22,266 --> 00:54:26,856

It's, it's their charge at the end of

a project to make sure that happens.

00:54:27,186 --> 00:54:30,066

And my next question obviously

is, well, does that happen

00:54:31,746 --> 00:54:32,376

Randall Stevens: Of course not.

00:54:32,796 --> 00:54:35,946

Evan Troxel: Because like, what

hooks does a digital practice

00:54:35,946 --> 00:54:37,566

have into a project manager?

00:54:37,626 --> 00:54:38,436

Like Zero?

00:54:38,496 --> 00:54:39,666

They don't have any, right?

00:54:39,666 --> 00:54:43,416

And so there's no responsibility

between those two.

00:54:43,911 --> 00:54:45,831

Different entities inside of a firm.

00:54:46,041 --> 00:54:47,031

It's a standard.

00:54:47,031 --> 00:54:50,001

And if the firm says it,

but does it actually happen?

00:54:50,391 --> 00:54:52,851

Um, and, and the answer's most likely no.

00:54:52,851 --> 00:54:54,411

Like it probably still doesn't happen.

00:54:54,411 --> 00:54:55,911

But it's because why?

00:54:55,911 --> 00:54:57,561

Because what if litigation?

00:54:57,561 --> 00:55:00,171

What if we want to use that

information on the next project?

00:55:00,171 --> 00:55:01,191

What if, what if, what if?

00:55:01,401 --> 00:55:03,711

And of course, like that

would all be useful.

00:55:03,771 --> 00:55:07,911

And at the same time, like there's so much

information and at the end of the project,

00:55:08,241 --> 00:55:12,801

do they actually build in the time for

somebody to be able to do what seems

00:55:12,801 --> 00:55:14,661

to me like kind of a monumental task?

00:55:15,171 --> 00:55:16,581

I doubt it, right?

00:55:16,581 --> 00:55:17,751

So it's a tough

00:55:17,831 --> 00:55:20,196

Carl Veillette: and, and

I've heard that many times.

00:55:20,196 --> 00:55:24,846

And you know, we've got customers

knocking on our door, you know.

00:55:25,161 --> 00:55:27,261

With that exact same problem right now.

00:55:27,411 --> 00:55:29,301

Now their litigation goes up, right.

00:55:29,301 --> 00:55:30,921

They've got more lawsuits.

00:55:30,981 --> 00:55:32,151

They, they lose the case.

00:55:32,181 --> 00:55:33,561

'cause they don't have the evidences.

00:55:33,561 --> 00:55:33,981

Right.

00:55:34,581 --> 00:55:37,431

And so they start to look

into is oh 9,001, right?

00:55:37,431 --> 00:55:41,001

They start to look at structured

their processes and all of that.

00:55:41,001 --> 00:55:44,766

And then, you know, like, you know,

at the end what they're like, what

00:55:44,766 --> 00:55:47,871

they're, what they're seeking is like,

is there something like an information

00:55:47,871 --> 00:55:52,341

backbone or a platform that can help

us restructure our processes, how we

00:55:52,341 --> 00:55:54,411

collect data, how we archive data.

00:55:54,801 --> 00:55:58,641

And, and that's often, you

know, like how we, um, how we

00:55:58,641 --> 00:55:59,901

get in touch with those folks.

00:55:59,901 --> 00:56:00,201

Right?

00:56:00,531 --> 00:56:03,321

We, we got some customers that

were referred to us by lawyers,

00:56:03,921 --> 00:56:05,121

which is kind of strange.

00:56:05,271 --> 00:56:06,141

It's a funny story.

00:56:06,881 --> 00:56:07,661

Randall Stevens: Put 'em on the payroll.

00:56:08,211 --> 00:56:08,501

Yeah.

00:56:08,586 --> 00:56:10,986

I was gonna, uh, when I,

when I first started here.

00:56:11,826 --> 00:56:14,406

Started working on, you

know, what became a veil.

00:56:14,916 --> 00:56:17,586

I was always trying to like

quantify or how to figure out

00:56:17,586 --> 00:56:21,516

how to describe the scale of the

problem of this, you know, data.

00:56:21,576 --> 00:56:25,656

And one of the things that I've said

for years that I was like, if you

00:56:25,656 --> 00:56:30,666

think about every piece of software

that we're all using, you know, we're

00:56:30,666 --> 00:56:32,136

all like information workers now.

00:56:32,136 --> 00:56:34,326

We're all sitting behind a

keyboard with some piece of

00:56:34,326 --> 00:56:35,706

software producing information.

00:56:36,456 --> 00:56:40,536

And the way I would always describe

the scale of the problem, at least

00:56:40,536 --> 00:56:44,196

a part of the problem is every

piece of software that has a file

00:56:44,226 --> 00:56:49,806

open and a file saved dialogue

requires a human to now do something.

00:56:50,406 --> 00:56:51,216

Where is it?

00:56:51,816 --> 00:56:54,306

And now that I've produced

it, where should I put it?

00:56:54,396 --> 00:56:58,416

And that's what got me thinking about

this idea of like, like a route.

00:56:58,476 --> 00:57:02,076

Like you would never ask your router where

those bits should go on your network.

00:57:02,076 --> 00:57:02,316

So I'm

00:57:02,676 --> 00:57:03,666

Evan Troxel: Look at a phone, right?

00:57:03,666 --> 00:57:06,276

I mean, at the phone there's barely a file

00:57:06,696 --> 00:57:09,156

Randall Stevens: It was the first

one to, you know, that it was, that

00:57:09,156 --> 00:57:13,176

was our first little bit of a taste

and go, you know, Google Drive was

00:57:13,176 --> 00:57:14,946

at least, but even there, right.

00:57:14,946 --> 00:57:19,446

I still, we still have basically folders

and places that we put stuff in those

00:57:19,686 --> 00:57:20,376

Evan Troxel: but they hide it.

00:57:20,496 --> 00:57:21,486

They obscure that stuff

00:57:21,546 --> 00:57:21,966

Randall Stevens: Yeah.

00:57:21,966 --> 00:57:23,526

It, it's obscured behind there.

00:57:23,526 --> 00:57:27,696

But I, I, you know, I do, I'll

keep, i'll, I'm, we're trying

00:57:27,696 --> 00:57:28,956

to solve some of these problems.

00:57:28,956 --> 00:57:32,676

I'm sure Carl, y'all are in the same,

but I think this, we gotta keep pushing

00:57:32,676 --> 00:57:36,966

on this, uh, that, that there needs to

be this metadata, you know, I don't know

00:57:36,966 --> 00:57:41,731

if those can become standards or, uh,

you know, defacto standards around how

00:57:41,796 --> 00:57:45,846

to put the right contextual information

wrappers on this stuff so that these

00:57:45,846 --> 00:57:49,146

systems, you know, this is part of what

we were trying to do with this tech stack

00:57:49,146 --> 00:57:54,156

survey too, which was at the Confluence

events to talk about these systems.

00:57:54,711 --> 00:57:57,951

We've got so many of them now and they're

not necessarily talking to each other.

00:57:57,951 --> 00:58:01,041

So I'm glad to hear, you know, you all

continue to work on your own connectors

00:58:01,041 --> 00:58:04,881

and that We did a, we did a little

webinar yesterday where I had some,

00:58:05,001 --> 00:58:06,471

some people on talking about this.

00:58:06,471 --> 00:58:12,021

And you know, part of it was, you know,

framed it as you're using these, you've

00:58:12,021 --> 00:58:17,721

decided to use these tools to accomplish

certain goals and tasks, and then you've

00:58:17,721 --> 00:58:20,181

got another tool to do another task.

00:58:20,421 --> 00:58:24,081

And then you have to find a third tool

that's trying to help you solve the

00:58:24,081 --> 00:58:26,781

problem of bridging, because those

two systems don't talk to each other.

00:58:26,781 --> 00:58:31,431

So can we get back to like, can we just

have better exchange of information

00:58:31,431 --> 00:58:35,811

between the core platforms and

systems that are in use and then maybe

00:58:35,811 --> 00:58:41,121

that'll absolve the need for another

tool that trying to fill those gaps.

00:58:41,121 --> 00:58:43,701

So I was referring to it as a seam.

00:58:43,701 --> 00:58:44,211

It's like.

00:58:44,766 --> 00:58:48,366

You can go, you, you, you choose the

tool 'cause you think it has value and

00:58:48,366 --> 00:58:52,686

it's gonna bring value, but it breaks

down at the seams where we can't get

00:58:52,686 --> 00:58:54,006

these things talking to each other.

00:58:54,006 --> 00:58:57,846

And that's where you pay the, you pay

the cost of like, uh, the friction

00:58:57,846 --> 00:59:01,836

of, of getting that data moved back

and forth, or knowing that if it's

00:59:01,836 --> 00:59:03,426

the right version of the data that

00:59:03,491 --> 00:59:03,821

Carl Veillette: Yeah.

00:59:04,091 --> 00:59:07,811

And, and it's the APIs and it's a various

system and it's also the native files.

00:59:08,321 --> 00:59:09,941

We're just talking about that, right?

00:59:09,941 --> 00:59:13,901

Like you think about like native files

format and that create additional

00:59:13,901 --> 00:59:16,031

barriers for consuming information, right?

00:59:16,866 --> 00:59:20,136

Uh, that's the reason I, I'm allergic

to, uh, you know, native file formats

00:59:20,136 --> 00:59:23,166

in a way that, you know, we've always

created, like, created like, like,

00:59:23,166 --> 00:59:27,756

um, you know, bridges and created

open APIs, and we've got those

00:59:27,876 --> 00:59:29,976

indexer with our secret sauce, right?

00:59:30,036 --> 00:59:33,216

Uh, which is basically our way to

index native file formats of the

00:59:33,216 --> 00:59:36,426

industry, like CAD files, Revit

files, things like that, right?

00:59:36,426 --> 00:59:41,646

So we, we kind of index all of that, but

the, the goal there is, has always been

00:59:41,766 --> 00:59:43,896

to free the data from the files, right?

00:59:44,286 --> 00:59:45,156

Um,

00:59:45,346 --> 00:59:46,346

Randall Stevens: 'em in

a database somewhere.

00:59:46,626 --> 00:59:47,556

Carl Veillette: yeah, exactly.

00:59:47,616 --> 00:59:51,936

So that, that's kind of like the, you

know, the reason why people call us the

00:59:51,936 --> 00:59:55,866

Google search of construction projects

often, like, you know, I, I would say if

00:59:55,866 --> 00:59:59,886

you ask most of our Newforma customers,

the common thing that would come up

00:59:59,886 --> 01:00:01,776

is newforma as a search engine, right?

01:00:01,956 --> 01:00:06,966

And so that, that's kind of a,

it's a strange category to be in

01:00:07,206 --> 01:00:08,976

as a software vendor in this space.

01:00:09,141 --> 01:00:10,941

Evan Troxel: That's what goes

on the survey, Randall, they,

01:00:10,941 --> 01:00:12,141

they go in the search category,

01:00:12,171 --> 01:00:12,681

Randall Stevens: Right, right.

01:00:12,696 --> 01:00:13,086

Carl Veillette: yeah.

01:00:13,671 --> 01:00:14,331

Evan Troxel: search engine.

01:00:14,673 --> 01:00:14,841

Carl Veillette: Yeah.

01:00:14,901 --> 01:00:17,331

Randall Stevens: Well, it'll be

interesting and you know, I'll, I'll

01:00:17,331 --> 01:00:21,171

use, you know, your all's help and

anybody's help that wants to help to

01:00:21,291 --> 01:00:25,911

think through what's the better, you

know, and like I said, it's, you can

01:00:25,911 --> 01:00:29,451

live in multiple categories, so it kind

of hard do you put everybody in every

01:00:29,451 --> 01:00:32,901

category that they might be in That way

you're at least collecting, you know,

01:00:32,901 --> 01:00:34,401

or they thought of, of being in that.

01:00:34,491 --> 01:00:38,181

Uh, but, uh, anyway, we'll, we'll,

we will figure it out as we go.

01:00:38,211 --> 01:00:41,781

I think it'll be, uh, I'm, look, I'm

already looking forward to, at the