When AI adoption stalls inside a contracting firm, leadership tends to reach for familiar explanations. The team is resistant to change. The older employees are not comfortable with technology. People are too busy to learn something new. These explanations are not entirely wrong, but they tend to miss the more specific and more fixable reasons that people do not engage with tools that could genuinely help them.
The honest answer is usually simpler and less flattering to how the introduction was handled.
People do not adopt tools they do not understand in the context of their own work. Not tools they do not understand technically, because most AI platforms require no technical knowledge to use, but tools they do not understand in terms of what they are supposed to do with them on a Tuesday afternoon when they have three things due and a subcontractor on hold. Abstract demonstrations of AI capability do not answer the question that every person in a training room is quietly asking, which is where exactly does this fit into what I actually do every day.
When that question does not get answered clearly, people leave the training with a general sense that AI is interesting and no real plan for using it. The tool stays unopened. The habit never forms. And six months later leadership wonders why nobody is using the platforms the firm invested in.
This plays out differently depending on the role. An estimator who watched a demo of AI summarizing a contract document walks away thinking that seems useful, but their day is built around takeoffs, subcontractor calls, and bid deadlines. Unless someone showed them specifically how AI fits into that workflow, with a real scope document and a real bid scenario, the connection never gets made concrete enough to act on. A project manager who saw AI generate a professional email from rough notes thinks that could save me time, but they have never actually tried it under real deadline pressure, and the first time they sit down to do it the friction of figuring out the prompt feels like more work than just writing the email themselves.
The gap is almost always between demonstration and application. Firms that close that gap show people how to use the tools inside the specific tasks they are responsible for, not in hypothetical scenarios but in the actual work sitting on their desk. That requires more preparation than a general overview, but it produces a completely different outcome. When someone uses AI on a real task and sees it produce something useful, the adoption decision is essentially made. When they only ever see it work on someone else’s example, it stays theoretical.
There is a second reason adoption stalls that is less about the training and more about what comes after it. People need permission to be imperfect while they are learning. AI use involves trial and error, prompts that do not work, outputs that need significant editing, approaches that seem promising and turn out not to be. In firms where people feel pressure to be immediately productive with every tool they pick up, that learning curve feels like a risk. It is easier to stick with the known workflow than to experiment with something new when there is no explicit space for figuring it out.
Leadership creates that space or it does not. Telling the team that AI is a priority while holding them to the same output expectations during the learning period sends a mixed signal that most people resolve by quietly deprioritizing the new tool. Firms that see real adoption give people enough room to learn without penalty, which does not require a formal policy. It requires a clear and genuine signal from the people running the firm that figuring this out is worth some short-term friction.
The third reason is that people do not see anyone above them using it. AI adoption in contracting firms tends to move from the middle out, driven by individuals who figured something out on their own. What accelerates it is when the principal or operations leader or whoever carries authority in the firm is visibly engaged with the tools themselves. Not as a technology enthusiast, but as a business leader who uses AI in the course of running the business and is open about that. That signal matters more than any training program, because it tells the team that this is real, that leadership is not asking them to do something they are not willing to do themselves, and that the firm is genuinely moving in this direction.
None of these are insurmountable problems. They are predictable ones, and firms that understand them can address them directly rather than waiting for adoption to happen on its own.
The team is not the obstacle. The conditions are. Fix the conditions and the team will follow.