AI is getting smarter every month. Is anyone at your business watching?

Picture two civil engineering firms in Montgomery County, evenly matched. Same licenses, same reputation, competing for the same site development contract. One turns in a proposal done the way it's always been done. The other turns in a more thorough one in half the time, with drainage and cost-estimating detail the first firm's team usually doesn't have bandwidth to pull together before a deadline. The second firm isn't smarter or more experienced. It quietly folded AI into how it drafts proposals and runs preliminary calculations months earlier. The first firm was still treating it as something to look into eventually.

Same engineers, same licenses, same market. The difference is which firm has someone actually watching what AI can now do for their work.

The gap isn't the technology. It's attention.

Large language models are not improving on a normal product timeline. What these models could barely do in 2024, they now do reliably, and the curve hasn't flattened. If anything it's steepened. That's good news in theory: whatever a firm needs is either already possible or will be within a year. In practice, it means the right tool for estimating, proposal writing, or plan review changes every few months, and last quarter's limitation is often already solved.

A principal or PM running the firm has no reasonable way to track that curve on top of running the firm. That's not a knock on them. It's a second full-time job stacked on the one they already have.

What an AI operator actually does, day to day

This is where "hire a consultant for a workshop" undersells it. A good AI operator isn't a one-time installer. They're closer to a standing function, the way a bookkeeper or an IT contractor is:

• Tracks the landscape, not just your tools. Knows when a new model release changes what's possible for proposal drafting, take-offs, or plan review, and knows just as often when it doesn't and it's safe to ignore the hype.
• Translates capability into operations. "This model can now reason over spreadsheets and drawings together" is meaningless to a principal engineer. "This means your next bid can include a cost breakdown your team didn't have time to build by hand" is not.
• Protects against expensive mistakes. Overpaying for a bloated platform, feeding client or site data into a tool that mishandles it, or building a workflow around a feature that gets deprecated next quarter.
• Re-evaluates on a cadence. What was the right call in January isn't automatically still right in July, given how fast the underlying models move.

None of that requires the firm's engineers to become AI experts. It requires someone technical paying attention on their behalf.

The full-potential case

The firms that get the most out of AI over the next few years won't be the ones that adopted earliest or spent the most. They'll be the ones with someone dedicated to continuously matching AI capability to how they actually win and deliver work, quietly and in the background, without the principals having to become AI experts themselves. That's the gap between a firm that owns a few AI tools and one that's built to compound the advantage as the technology keeps improving.

If you're not sure whether your current proposals and workflows still reflect what's possible today, that's exactly the kind of question a second set of expert eyes answers quickly.