# Disrupting the $15 Trillion Construction Industry with Autonomous Agents: Dr. Sarah Buchner AI Engineer · 2024-08-22 Dr. Sarah Buchner, Founder & CEO of Trunk Tools, argues that vertical AI agents are the future, targeting the $15 trillion construction industry where 10% rework costs $1.5 trillion annually. Her company deploys an agent every 45 days, starting with TrunkText, a RAG-powered Q&A tool that saves field professionals 1-2 hours daily by accessing up to 3.6 million pages of documentation per skyscraper. The episode reveals how their system uncovers costly data discrepancies—such as a door needing power-actuated hardware that contradicts specs—and automatically generates RFIs to resolve them. Buchner insists RAG is commoditized, but keeping humans at the center with an army of agents solving real-world problems is the true impact. The talk highlights Trunk Tools' brain behind construction, backed by Sequoia, Accel, and others, and notes the company is hiring as it scales. ## Questions this episode answers ### What is the annual cost of rework in the construction industry due to data discrepancies? Dr. Sarah Buchner of Trunk Tools explains that construction is a $15 trillion per year industry, and 10% of that—$1.5 trillion annually—is rework caused by data discrepancies in the massive documentation that workers are unaware of. This rework results from mistakes in the 3.6 million pages of contracts, blueprints, and other files for a single skyscraper, and Trunk Tools aims to eliminate these costs using AI. [2:12](https://aie.addtry.com/a48fb307-0e3a-4293-b1da-da1fd60acbb9?t=132000) ### How many pages of documentation does Trunk Tools handle for a single skyscraper? Dr. Sarah Buchner says that for a skyscraper being built in New York City using Trunk Tools' software, the company was given access to 3.6 million pages of documentation, including contracts, blueprints, schedules, and RFIs. If printed, this stack of paper would be three times as high as the building itself, highlighting the unstructured data mess that Trunk Tools' RAG system digests. [0:54](https://aie.addtry.com/a48fb307-0e3a-4293-b1da-da1fd60acbb9?t=54000) ### How does Trunk Tools' AI agent handle data discrepancies found in construction documents? After Trunk Tools' RAG system finds a data discrepancy in the 3.6 million pages of documentation, an AI agent automatically creates an RFI (Request for Information) to resolve the issue. This agent then solves the discrepancy so that the human can go back to doing their job, instead of being a victim of bureaucracy. [4:03](https://aie.addtry.com/a48fb307-0e3a-4293-b1da-da1fd60acbb9?t=243000) ## Key moments - **[0:00] Intro** - **[0:43] Data Overload** - [1:10] A single New York skyscraper amassed 3.6 million pages of construction documentation, a stack three times the building's height. - **[2:12] Rework Crisis** - [2:12] 10% of the $15 trillion construction industry is rework, costing $1.5 trillion annually due to data discrepancies. - **[2:28] RAG Brain** - **[2:57] Live Demo** - [3:47] TrunkText's Q&A agent uncovered a costly mistake in a skyscraper's 3.6 million pages by surfacing semantic discrepancies. - **[4:30] Vertical Future** - [4:30] RAG is commoditized; the future is vertical AI agents solving real-world problems, says Dr. Buchner. - **[5:04] Join Us** ## Speakers - **Dr. Sarah Buchner** (guest) ## Topics RAG & Retrieval, Autonomous Agents ## Mentioned Trunk Tools (company), Copilot (product) ## Transcript ### Intro **Dr. Sarah Buchner** [0:14] Welcome, everybody, to this little afternoon session of mine. I'll keep it very short, but this is going to be the most exciting presentation of the day because I'm talking to you about construction. **Guest** [0:23] Whoo! **Dr. Sarah Buchner** [0:24] Thank you. We are Trunk Tools. Trunk Tools is the leading generative AI provider for the construction industry, which probably none of you have ever thought about it, but you are consuming construction every single day. You're sitting in itright now, you're driving in it on your day home, on your ride home, and it is the most exciting industry in the world. ### Data Overload **Dr. Sarah Buchner** [0:43] I'm going to level set the stage here a little bit. This is a building that is being built in New York Cityright now, and it's using our software. We asked this building to give us access to all their documentation, okay? Raise your hand if you think they gave us access to more than 1,000 pages. Come on, don't be shy. This is not an engineering conference. Keep your hand up if you think there's more than 10,000 pages of documentation, 100,000 pages of documentation, a million pages of documentation to build a single skyscraper. They gave us access to a total of 3.6 million pages of contract. Okay, 3.6 million pages of pictures and blueprints and schedules and RFIs and all of this stuff. If you were to print that, a stack of paper three times as high as the building itself. That's why I'm super excited about this industry, and that's why this industry is literally built for RAG. That's why they put me up on this RAG stage here. Why do we have such a problem in construction doing this? There are massive amounts of people making decisions every single day, and they're documenting the heck out of this because it's one of the most litigious industries that you will ever see. Similar to software development,right? Like, I have a PhD in AI and construction, so I kind of know what I'm talking about, but I'm not an expert. That's why we have better people on my team. But similar to software development,right, you have lots of people writing code, but once you merge it together, it actually breaks and you have bugs. In construction, that happens once you are in the field and you're trying to pour the concrete or you're trying to squeeze a pipe into a hole and it just doesn't fit. Construction is $15 trillion a year. Trillion, with a T. 10% of it is rework. 10% is because there's data discrepancies in this pile of data that the people don't know about it. $1.5 trillion every single year. And that's what we are solving. ### Rework Crisis **Dr. Sarah Buchner** [2:28] We are putting all of this knowledge in one place, which is what we call the brain behind construction, obviously a RAG product that goes in, perfectly digests and memorizes all of these construction-specific file types. Yes, we built RAG specifically for a vertical. ### RAG Brain **Dr. Sarah Buchner** [2:43] This vertical, however, is 10% of GDP. Once we have built this, we are deploying on top of this brain an army of AI-based agents. Agents are more or less intelligent, and we can dive into this if you come to our booth a little bit later. ### Live Demo **Dr. Sarah Buchner** [2:57] I'll show you one of these agents live in practice in the field. Okay, this is a traditional Copilot. You have all seen this a million times. You ask a question about your 3.6 million pages of construction data. Within a few seconds, you get an answer, and you obviously also want the source to the original file document. Used on billions of the construction volumeright now, I'm showing you an example from the actual construction side. That's door number 2103, number 2, needs power-actuated hardware. Yes, a very important question on the construction side. The way this works for us, you throw this question against the system via text message or via web app. Within a few seconds, we give you the enterprise search results, and obviously we also give you a semantic answer to your actual question. Interestingly enough, for anybody who actually wants to read what's up there, there are discrepancies in the answer. The answer itself shows that in the 3.6 million pages of documentation, actually somebody made a mistake, and that mistake is going to be very, very expensive. To give you a sense of the unstructured mess of data we are dealing with in construction, let me open one of these files. Beautiful. That's how we structure documentation in construction. Three feet wide tables with a ton of numbers, unusable for any human. That's why we're throwing AI against it. Obviously, I was talking about data discrepancies. Instead of stopping where we have the data discrepancies, why don't you use an agent that is just actually creating an RFI and solving the discrepancy for the human so that the human can go back doing their job and not being a victim of bureaucracy. ### Vertical Future **Dr. Sarah Buchner** [4:30] I'm on a RAG stage, but personally, I believe that RAG is completely commoditized. So what am I doing here? I'm here to tell you, for whoever is building RAG, that the future of RAG is actually keeping the human in the center and augmenting the human with an army of agents that is solving real-life problems. Because if you actually want to have an impact in real life, then in my opinion, vertical agents solving real-world problems are the future. If you're interested in joining us on that path and being in the sexiest industry in the world, come join us at our booth in the expo hall. ### Join Us **Dr. Sarah Buchner** [5:04] We're called Trunk Tools. Here's the QR code. We are hiring left andright, and we're very excited to be dominating one of the most impactful industries in the world. Thank you. --- This library is powered by PodHood (https://podhood.com), the podcast website platform.