AI augmentation, not replacement: an honest take from inside a 12-person crew

A 12-person remediation crew in the Inner West adopted AI-assisted quoting eighteen months ago. The vendor pitch said one estimator could do the work of three. What actually happened: they kept their senior estimator, hired a second, and more than doubled the quotes they could turn around. The senior is happier. The crew is busier. The vendor was right about the productivity gain and wrong about how it would get spent.

01What estimators actually do all week

A senior estimator on a busy crew spends roughly 60 percent of their week on data entry, formatting, and re-keying. The other 40 percent is the work that justifies the salary: judgement calls, supplier negotiation, risk assessment, the conversation with the strata manager that turns a $90k job into a $110k job because both sides understand what is included.

Almost everybody, asked privately, describes the 60 percent as the worst part of their week. They tolerate it because nobody has shown them a way around it that does not introduce errors.

02AI eats the 60 percent

Transcribing voice notes, looking up unit rates, formatting line items, attaching compliance references: this is exactly the work the AI does well. It also happens to be the part of the job that produces the most Friday-night mistakes. Removing it from human hands is a productivity win and a quality win at the same time.

The 60 percent does not disappear. It compresses. What was four hours becomes thirty minutes of review. The estimator goes from typing to checking.

03The 40 percent expands into the space

When data entry collapses, the time saved does not vanish. It flows into the work that matters. More site visits. More careful pricing on the trades that drive margin. More follow-up calls with strata managers, building certifiers, suppliers. The work becomes more like construction expertise and less like data warehousing.

This is the bit the vendor pitches miss. They sell productivity. The crew that adopts properly buys quality. The number of quotes goes up. The quality of every quote goes up. The senior estimator stops looking for a way out of the job, because the job is now closer to what they signed up for.

04Hiring decisions look different on the other side

The crew that doubled quote volume now needed a second estimator. Pre-AI they could not justify the salary, because adding an estimator only delivered another 60-percent-typing-and-40-percent-thinking person. Post-AI, a new hire is 90 percent thinking and 10 percent reviewing. The economics of the second estimator look entirely different.

This pattern is showing up consistently across the crews we observe. AI does not replace estimators. It makes hiring another one commercially obvious. The crews that pretend the technology will let them downsize either lose people first and discover the productivity gain is smaller without senior judgement, or never quite trust the AI enough to act on the headcount call.