This story was originally published by Jessica Brown on the Bluebeam Blog.

Construction teams face a daily flood of data—plan documents, RFIs, contracts and submittals—all essential but often buried in complex document management systems.

What if finding answers was as simple as asking a question?

Sarah Buchner, founder and CEO of Trunk Tools, set out to make this vision a reality. “The biggest problem in construction is that we have a lot of data and we’re hardly doing anything with it because it’s so unstructured and difficult to analyze,” Buchner said.

Her company uses generative AI to provide a solution: a conversational, chat-based tool that lets project managers and superintendents access the information they need immediately. “You ask a question and get an answer, with links to original source documents. Click, and you get the original artifact.”

Inside TrunkText: How Generative AI Powers Smarter Construction Workflows

TrunkText is Trunk Tools’ chat agent that construction workers use to query project data. Powered by a large-language model (LLM), it can even automate workflows like comparing a submittal against project specifications.

“Instead of having a human sitting there comparing the two documents, it’s something AI can do,” Buchner said. “There are so many workflows in our day-to-day life that we just don’t need humans to do anymore.”

Real-World Impact: How TrunkText Streamlined the Baird Center Expansion Project

In 2021, Gilbane Building Co. began work on the $456 million Baird Center Expansion in Milwaukee, Wisconsin, a massive project involving more than 33 GB of data across approximately 21,000 documents. As deadlines approached, sifting through this mountain of data consumed precious time.

How Large-Language Models (LLMs) Turn Construction Data into Answers

Construction projects generate massive amounts of unstructured data—think plan documents, RFIs, submittals, contracts and daily reports.

The problem: Much of this information lives in separate files, formats and platforms, making it tough to access and analyze when decisions need to be made quickly.

This is where large-language models (LLMs) come in. LLMs are a type of artificial intelligence trained on vast amounts of text data. They excel at understanding language patterns, context and meaning—skills that make them perfect for simplifying complex construction data.

Here’s how LLMs work in construction:

• Data Processing: When documents like contracts and submittals are uploaded into an AI-powered platform like TrunkText, the LLM scans and indexes their contents, turning them into searchable data.
• Semantic Understanding: Unlike basic search engines that rely on keywords, LLMs understand the meaning behind queries. This means that if a project manager asks, “What’s the maximum load capacity for beam X?” the AI can find the relevant spec—even if the exact words aren’t in the document.
• Contextual Answers: LLMs provide context-rich answers by linking directly to the source documents. Users don’t just get a response; they get the evidence behind it.
• Automating Repetitive Tasks: Beyond answering questions, LLMs can also handle tasks like comparing specifications or generating summaries, reducing the need for manual work.

By turning unstructured project data into clear, actionable insights, LLMs can help construction teams save time, avoid costly mistakes and focus on getting the job done right. In a high-stakes industry where every minute counts, that’s a game changer.

To address this, Gilbane piloted Trunk Tools during the project’s final sprint in January 2024. Once documents were uploaded, the AI processed the data, enabling field workers to quickly retrieve information like plan document details or change orders.

“More than 30 minutes is often saved on questions asked in the field,” said Andrew Roy, superintendent on the Gilbane project. “It’s also positively benefiting the speed to response for the person asking the superintendent a question.”

By project completion in May 2024, the AI had answered nearly 250 questions, saving 20-40 minutes per query—equating to an estimated $100,000 or more per month in avoided rework costs. Critically, 87% of the answers were verified as accurate, giving teams confidence in its reliability.

From Job Sites to AI Innovation: Buchner’s Journey in Construction Tech

Buchner’s passion for construction began in Austria, where her father, a carpenter, often took her to job sites. “I started working on my first job when I was 12, and liked it,” she said. After nearly a decade in construction, she pursued advanced degrees in civil engineering, data science and business.

Top 3 Challenges AI Is Solving in Construction

1. Data Overload: Construction projects generate thousands of documents, from plan documents to contracts. AI-powered tools organize and index this data, making it searchable and manageable.
2. Slow Information Retrieval: Finding critical project details can take hours. AI chatbots like TrunkText provide instant answers, linking directly to source documents.
3. Manual Document Comparisons: Tasks like checking submittals against specifications are time-consuming. AI automates these comparisons, reducing human error and speeding up approvals.

By addressing these challenges, AI is transforming how construction teams work—saving time, cutting costs and boosting efficiency.

While working on her Ph.D., Buchner encountered firsthand the difficulties of analyzing unstructured construction data. “I had to raise money to clean the data so I could do analytics,” she said. This experience highlighted the urgent need for better data management and access in the industry. When generative AI emerged, Buchner realized it could revolutionize construction workflows.

What’s Next: How AI Is Shaping the Future of Construction

Since founding Trunk Tools in 2021, Buchner has focused on creating AI tools that address real problems faced by field teams. The chat-based interface is intentionally intuitive, requiring minimal user training.

“Trained AI is really good at semantically connecting data with humans,” she said. “We’re used to talking to get the information we need. When more solution providers figure that out, we’re going to see a lot more chat-based tools like this.”

With tools like TrunkText, generative AI is no longer just a buzzword—it’s reshaping how construction teams work, saving time, cutting costs and unlocking the potential of data in the field.