Before, during, and what changed.
The before — a phone Tammy couldn't put down
Tammy runs a home health care agency in central North Carolina. The business is built on family referrals and word-of-mouth — which means calls don't follow office hours. Daughters call at 9 PM when Dad just got home from the hospital and they need someone to start care tomorrow morning. Sons call Saturday at 6 AM because Mom had a fall overnight.
For three years, Tammy answered every one of those calls personally. After-hours, weekends, sitting at dinner with her own family. The phone was always within arm's reach. She told us the worst part wasn't the calls she took — it was lying awake at 2 AM wondering which one she had missed and whether that family had already called a competitor.
The numbers backed up the worry. She tracked it for one month before we started: 38 calls hit voicemail. Of those 38, only 4 left messages. The other 34 were potential new admits — at roughly $3,800 each in monthly recurring revenue per client — gone.
The build — one week from "I'll try this" to live
Kickoff was a Wednesday morning. 35 minutes on Zoom. Tammy walked us through the top questions families ask, the difference between her three service tiers, her referral-partner workflow, and the one absolute rule: no AI ever schedules a same-day skilled-nursing visit without escalating to her first.
By Friday we had a working demo. Tammy spent the weekend calling it, trying every objection she'd heard over the years. She sent us 7 tuning notes. We closed them all by Monday afternoon. Tuesday she signed off on the go-live script. Wednesday her business line forwarded to the AI.
Total elapsed time from signup to "every call answered": 7 days. She didn't have to learn new software. Her existing phone number didn't change. Her referral partners didn't notice anything until they called and got a faster, friendlier answer than they were used to.
The after — sleep, revenue, and a calendar that fills itself
Month 1 was about learning what the AI could handle. Tammy escalated more calls to herself than she needed to, just because she didn't yet trust it. By month 2 she was letting the AI book 85% of new-client intake calls directly to her calendar. Her admit rate went up. Her after-hours call load went to zero.
The revenue number — $18K/month recovered — is calculated conservatively. We counted only the calls that previously went to voicemail and were demonstrably new admits in month 2 vs. her pre-launch baseline. Doesn't count the referral-partner relationships strengthened by faster response, doesn't count the time she got back to focus on running the business instead of answering the phone.
The thing she keeps mentioning, though, isn't the revenue. It's that she sleeps. She doesn't reach for the phone at 11 PM anymore. The phone got handed to someone who doesn't get tired.