My Favorite Moments from Episode 247 "It's on the Server Somewhere"
💬 Twenty-one lines worth keeping from TRXL 247. Fifteen voices on where the reasons behind a project's decisions end up, why shared meaning beats one standard, and what to do before an AI reads your drive.
| Arc 3 presenting sponsor | |
| Speckle makes design data actually usable across projects, so your team can trust it, own it, act on it, and make better decisions faster. | |
Every episode of this series, I go back through the full conversation and pull the lines I don't want you to miss. Here are 21 of them from Episode 247, “It's on the Server Somewhere.” Each one has who said it and a timestamp if you want to hear it in context.
On where the why goes
![]() | “Up to 80% of the data on that project, not data outside of that project, but data on that project, doesn't sit in those tools or systems. And so there's so much communication happening, so much other stuff that doesn't live in a project delivery platform, and that's where the decisions are made. That's where the why sits.” Aaron Vorwerk, Egnyte · ▶ 0:09:39 |
![]() | “Comparisons are easy. Why that was changed is a very different animal and is not represented in the contract deliverables at all. It's in the emails, it's in the Slack threads, the Teams chats, the digital whiteboards, beautiful places where things go to die.” Aaron Vorwerk, Egnyte · ▶ 0:11:38 |
![]() | “There's always one person at a firm that knows everything. They know where all the skeletons are in the closet. … But how do you capture that? That's tricky.” Ross Wagner, Pirros · ▶ 0:15:55 |
![]() | “There's such a lack of mentorship sometimes at a firm that's moving so fast that you just have to slow down and teach people and show what good looks like.” Ross Wagner, Pirros · ▶ 0:16:24 |
![]() | “It doesn't have to be perfect. It just needs to be written down. We just need to document the decision, have the discussion, do the back and forths so we can come back in the future, we can reference it, or we can learn from it.” Scott Reynolds, UpCodes · ▶ 1:18:23 |
On shared language and messy data
![]() | “You don't need everyone speaking the exact same language, but you need enough shared meaning for your data to move across different disciplines and life cycles.” Andy Rahden, Bentley Systems · ▶ 0:21:29 |
![]() | “Where it breaks is when you try to force a monolithic standard across every single discipline. That's when you get resistance, and that's when you start getting workarounds.” Andy Rahden, Bentley Systems · ▶ 0:22:03 |
![]() | “There is this false bill of goods maybe that was sold to everyone when it comes to BIM, and I think everyone visualized everyone in the project's going to be working in this collaborative space. One big file that's got absolutely everything in it.” Chris Razzell, dRofus · ▶ 0:50:47 |
![]() | “The space is king in Finch, while in almost all other software, the wall is the main element.” Jesper Wallgren, Finch · ▶ 0:37:35 |
![]() | “Garbage in, garbage out. You need to be careful around this. And to a certain extent, that stands true. But there's added nuance today. An AI agent can whiz through your inconsistencies in naming or in misclassification of parameters, and can fix those things in an instant.” Dimitrie Stefanescu, Speckle · ▶ 0:22:43 |
![]() | “There's a lot of people out there who say, “Well, our BIM's not good enough, and we really need to clean up our data so that we're ready for AI.” The AI doesn't care. Your data is whatever your data is. It's a cluttered garage.” Marty Rozmanith, Skema · ▶ 0:34:37 |
On starting before the structure is perfect
![]() | “I would caution any firm who is procrastinating or not starting until they know exactly how to structure it. I would say just go ahead and start structuring it because the answer is going to reveal itself as you're working and asking questions to your own data.” Ariel Cooke-Zamora, Bimbeats · ▶ 0:03:26 |
![]() | “These systems, they're going to hallucinate, and it's up to firms to, one, structure their data, and then two, define the skills that they're going to give the agents to interpret that data.” Ariel Cooke-Zamora, Bimbeats · ▶ 0:03:15 |
![]() | “Whenever you pair an agent and a simple old-fashioned database, a SQL database, it will always work.” Theodoros Galanos · ▶ 1:31:16 |
On where the data lives and who can see it
![]() | “The best information's at the edges of the network. It's the people doing the work that actually know.” Randall Stevens, AVAIL · ▶ 1:06:50 |
![]() | “Windows file folders were never designed to hold and capture anything except storing the file and then putting a name on a folder that gave you some small amount of contextual information to be able to know what's in that folder, and then it stops.” Randall Stevens, AVAIL · ▶ 1:07:57 |
![]() | “Governance has to come first. In our space, there's some folks that are pretty laissez-faire about this, that I think there will be bites. I would bet my house there's going to be some major breaches in the next 12 to 18 months because of this stuff.” Mike Soldan, Egnyte · ▶ 1:23:25 |
![]() | “Security through obfuscation. … It felt safe, but once that content's indexed and AI can find it contextually, it's going to surface it up really fast. So this issue of messy data just becomes an even bigger problem.” Mike Soldan, Egnyte · ▶ 1:26:31 |
On habits that make the data worth keeping
![]() | “A lot of the time people do things because it's fast, so we have to make using the correct standard faster.” Sasha Crotty, Autodesk · ▶ 1:13:35 |
![]() | “The benefit of me doing this slightly differently is just so much bigger than the pain of doing it slightly differently.” Matt Wash, Bimbeats · ▶ 1:15:28 |
![]() | “So you can go from the very simple, “Hey, you've just helped me do my timesheet,” to now organizationally, how do we deliver our projects based on actual time rather than the fictitious timesheet?” Matt Wash, Bimbeats · ▶ 1:17:14 |
😎
Win a pair of Ray-Ban Meta Smart Glasses courtesy of Graitec. Enter Here. (entering shares your contact info with Graitec)
That's the whole episode's worth of highlights. Full conversation's at trxl.co/247 if any of these made you want the rest.














