A mid-sized contracting company came to us believing their biggest operational problem was project management software. When we did the audit, we found something simpler — and far more expensive.

The owner had been running a general contracting business for eleven years. 18 employees. Residential and light commercial work. Good reputation, solid referral volume, consistently busy. He came to Chatpliance because a friend had mentioned AI and he was curious whether it could help with project scheduling.

We started where we always start: with the audit. Not the technology — the workflow.

What the Audit Found

The company received an average of 140 inbound contacts per month. I asked his office manager to keep a log for two weeks, categorising each contact by type. Eight question types accounted for 84% of all inbound contacts:

• Do you do this type of project?
• What is your approximate price range for X?
• Are you taking on new projects — what is your availability?
• What is the status of my current project?
• Can I get a quote — how does the quoting process work?
• Are you licensed and insured in my area?
• What materials do you use?
• What is your payment schedule?

Eight questions. 84% of all contact volume. His office manager was spending approximately 22 hours per month answering variations of these eight questions. At her fully-loaded rate, that was approximately $3,900 per month — $47,000 per year — in labour cost for information that was either already on their website or answerable in a single sentence.

$47,000 per year. Not on complex client management. Not on anything that required judgment. On answering the same eight questions, 140 times a month, for eleven years.

What We Built

We trained CynAI on the company’s service scope, pricing bands, project intake process, licensing details, material policies, and payment terms. Clear escalation rules: anything involving a specific commitment, a sensitive situation, or a new lead ready to book a site visit goes to a human immediately. Everything else — the eight question types — the AI handles.

What Happened After

In month one, the AI handled 118 of 147 contacts without escalation. The office manager reviewed the conversations weekly. By week five, she was flagging one or two edge cases per week. The 22 hours per month of routine information delivery became 4 hours of reviewing AI conversations and handling genuinely complex matters.

The owner told me three months later that the most significant change was not the cost saving — it was that his office manager had stopped being stressed. She had been on the edge of leaving. The job had felt like an endless queue of the same conversations. It no longer felt that way.