By Elliot Christiansen, Sr. Vice President of Operations - Cleveland Construction, Inc., United States
For most of my career, construction technology has been treated as something separate from construction operations. It was the new platform, the new app, the new dashboard, or the new tool someone wanted the field to use. That approach does not work.
Technology creates value in construction only when it is tied directly to how we build. It has to reduce friction for the people doing the work. It has to help project teams make decisions faster, improve consistency, and manage risk before it becomes cost, schedule, or client impact.
Over the last 12 months, my focus at Cleveland Construction has been moving AI out of the idea stage and into the actual workflows our project teams use every day. I have not approached AI as a trend or a side experiment. I have approached it as an operations strategy, focused on helping our teams build better, make faster decisions, reduce repetitive work, and expand capacity without growing headcount at the same pace as our workload.
Construction Has a Capacity Problem
The construction industry has talked about labor shortages for years, decades, but the issue is bigger than craft labor. We also have a capacity problem in project management, estimating, safety, compliance, document control, closeout, and operations.
Schedules are tighter, owners expect faster decisions, material lead times remain a constant risk, and design teams and trade partners are stretched. Strong project teams spend too much time looking for information that already exists in the project record.
That is where AI can create real value, but only if it is implemented in a practical way. The lesson for me has been simple. AI adoption only works when it helps the field and project teams solve problems they already have. It needs to be tied to the systems where the work already happens, understand the documents and project context, and earn trust through accuracy, transparency, and practical value.
Start with the System of Record
One of the most important lessons from our work is that AI is more valuable when it is available directly inside (or directly integrates with) the project system of record. For Cleveland Construction, Autodesk Construction Cloud (Autodesk Forma) is a major part of that foundation.
The value is not just document storage. It is a connected environment where drawings, specifications, RFIs, submittals, cost information, forms, schedules, and project communications can live together.
That matters because AI is only as useful as the information it can reach. A project team member should not have to leave the platform, download documents, upload them somewhere else, ask a question, and manually copy the result back into the project record. That creates version control issues, extra work, and risk.
Autodesk Assistant and AutoSpecs are examples of why embedded AI matters. Autodesk Assistant brings support into the workflow where the team is already working. AutoSpecs helps identify required submittals from the specifications and supports better submittal log development. These tools do not remove human review, but they give the team a better starting point and reduce missed requirements early in the project.
Submittals and RFIs Are Schedule Risks
The submittal process is one of the clearest examples of where AI can create operational value. In a typical workflow, it can take 15 to 20 days to receive an initial submittal from a subcontractor. The general contractor may spend 3 to 5 days on internal review. The architect may take another 10 to 15 days. If another consultant needs to review it, that can add roughly 10 more days.
If the submittal is rejected, the process starts over. That is not just paperwork. It is schedule risk. Rejected submittals can delay procurement, which can delay installation and downstream trades. Missed issues can be even more costly. If the wrong material is approved, purchased, delivered, and installed, the project may be dealing with rework late in construction.
TrunkSubmittal helps address that risk by comparing a submittal against the project specifications, drawings, prior RFIs, and related project data. It helps the team identify compliance issues before the submittal moves forward. A strong APM or PM still owns the review. AI strengthens that process by helping the team catch issues faster and return better comments to the subcontractor.
RFIs require the same discipline. Sometimes an RFI is submitted before the project team has fully checked the documents, prior RFIs, sketches, or meeting history. Sometimes the answer already exists. Sometimes the RFI is needed, but it is not written clearly enough.
Trunk RFI is being developed to help screen that process before an RFI is submitted. If the answer already exists, it can point the team to the proposed solution. If the RFI is needed, it can help draft the question and attach the right support. The goal is to reduce unnecessary RFIs and improve the quality of the RFIs that do get submitted.
Field Adoption Comes From Real Value
For me, the test is not whether an AI demo is impressive. The test is whether a superintendent, APM, or project manager uses it in the middle of real work because it saves time and lowers risk.
That is why tools like TrunkText matter. A superintendent or project manager should not have to spend 20 minutes digging through project documents to answer a question that can be found in two minutes. The time savings are important, but the bigger value is decision speed.
Construction is full of small decisions that add up. When teams can answer those questions faster, with better context, the project moves better.
Our work with OpenSpace, Disperse, and OpenSpace Track follows the same principle. For a national contractor, visibility is an operational challenge. Reality capture and progress tracking help us compare what is planned against what is actually happening in the field. That visibility supports better conversations, earlier intervention, and stronger accountability.
Data Governance Matters
AI adoption also has to be balanced with data protection. Construction companies handle contracts, pricing, HR information, legal documents, project records, and client data. We need to control where data lives, who can access it, and where it goes after it is used.
That is one of the reasons Box AI Studio and Box Hubs are important in our broader strategy.
Box gives us a way to apply AI to company content while keeping permissions, access, and
governance in place. We can protect the data while still making it more useful to the people who need it.
The ability to use different model families inside Box AI custom agents is also important. As OpenAI, Gemini, Llama, Claude, and other models improve, we do not want to be locked into one approach. We want the flexibility to use the right model for the right workflow while keeping our data inside a governed enterprise environment.
Vendor Partnerships Accelerate the Work
Another lesson from the last 12 months is that contractors cannot sit back and wait for perfect software. We need to work with the right technology partners and be willing to give regular, direct feedback, and expect excellence from them just as our clients expect excellence from us.
When we operate as a true development partner, we can influence product roadmaps, shape configurations, and help vendors solve the problems that actually exist in the field. That reduces the need for expensive custom programming. It speeds up implementation and training because the product is closer to the way the work actually happens. It also helps ensure new tools integrate with the rest of the stack instead of becoming another disconnected system.
That has been a major part of our strategy. Use existing platforms where possible. Push them to improve. Integrate them into the broader stack. Avoid one-off custom builds unless they are truly needed. Keep focusing on operational value.
In the age of AI, the only impossible feature request is the one you never bring to the table, which is why contractors must be active development partners, not passive software customers.
Conclusion
The right way to talk about AI in construction is not simply headcount reduction. The better framing is capacity expansion.
Can we do more work with the same number of people? Can we reduce the administrative burden on project teams? Can we help APMs and PMs spend less time chasing information and more time managing risk? Can we help superintendents get faster answers in the field? Can we help executives see across the portfolio more clearly?
That is the opportunity. AI should help builders build. It should reduce the tedious work that keeps strong people from focusing on the work that matters most. It should improve decision quality, protect schedules, reduce rework, improve compliance, and create better outcomes for clients.
The construction industry does not need more technology for the sake of technology. It needs better connected workflows, governed data, practical adoption, and tools that fit the way project teams actually operate.
Contractors do not make money by experimenting with AI, they make money by putting work in place, and the only AI worth scaling is AI that helps our teams build faster, safer, smarter, and with less friction.
