Illustrative case study — composite of a typical Pilot install. Riverside Dental of Austin is the synthetic business we use to test every receptionist on the leaderboard. Real-customer case studies will be published once we have post-launch results to share. We'd rather show you a realistic composite than fake testimonials.
Case study · Pilot · Dental
How Riverside Dental
stopped losing 14 calls a week.
Family dental practice on Riverside Drive, Austin. Two doctors, four staff, busy front desk. Their problem wasn't the patients walking in — it was the ones who called at 5:47pm and got voicemail. Here's the 7-day install that fixed it.
Tier
Pilot
Install time
7 days
One-time
$997
All-in monthly
~$240
Before
The bleeding.
Riverside Dental's phone rang ~80 times a week. Front desk caught about 66 of them. Fourteen rolled to voicemail — most of those callers never left a message and most of those didn't call back.
The math was painful once they ran it. A new patient is worth ~$850 in year-one revenue (cleaning, exam, X-rays, plus the average filling or two). Even at a 30% close rate on inbound calls, 14 missed calls/week meant ~5 lost new patients per month. That's $4,250 of new-patient revenue evaporating monthly because nobody picked up the phone.
They'd looked at hiring a part-time evening receptionist (~$2,200/mo loaded cost) and at a SaaS receptionist service ($1,800/year minimum plus per-minute fees). Both felt heavy for the actual problem.
7-day install timeline
What happened, day by day.
Day 0
Discovery & kickoff
30-minute call. Diagnosed the bleeding: ~14 missed calls a week (Twilio carrier logs from their existing line), translating to ~5 missed new-patient bookings/month. Agreed scope: AI receptionist on a parallel line, forwarded after 3 rings during business hours and 24/7 outside them.
Day 1
Knowledge base built
Hours, services, prices (cleaning $250, exam $250, X-rays $150, fillings from $300, Invisalign from $6,500), accepted insurance (Delta Dental, MetLife, Cigna, BCBS, Aetna), emergency protocol (capture name + callback, page on-call doctor, do not triage). Reviewed and approved by Dr. Reeves in 12 minutes.
Day 2
Stack provisioned
Synthflow account ($79/mo plan), Twilio number ($1.15/mo + $0.013/min usage), ElevenLabs voice (using 'Adriana' — warm, mid-Atlantic, picked from a 3-option blind A/B with the front-desk team). Anthropic API attached.
Day 3
Integration
Google Workspace OAuth (15 min). Calendar integration so Iris can read availability live. Airtable lead database wired up so every captured caller appears in a dashboard the office manager checks each morning.
Day 4
Voice + prompt tuning
System prompt written: 'You are Iris, the after-hours assistant at Riverside Dental of Austin.' Hard refuse rules on clinical advice, prescription questions, anything outside the published KB. Disclosure on every call: 'Hi, this is Iris, the AI assistant at Riverside Dental — how can I help?'
Day 5
Internal test pass
Ran the 5-scenario leaderboard test. Scored 32/40 (Iris install) — beat Synthflow's stock template (27/40) and got within 2 points of the Upfirst-ranked top performer. The one weak axis was 'unknown question handling' — Iris hedged on a question about teeth-whitening pricing that wasn't in the KB. Added a fallback rule, re-tested, axis went 3→5.
Day 6
Real-world test
Dr. Reeves called her own AI from her cell. Office manager called from hers. Hygienist tested an 'emergency' scenario (broken crown, in pain, weekend). All three reports back: 'sounded like our front desk on a good day.'
Day 7
Go live
Twilio forwarding rules enabled: ring desk 3x during business hours → roll to Iris; route directly to Iris outside business hours. First real call came in at 4:47pm CT — appointment booking for the following Tuesday. Captured cleanly.
What was installed
The technical stack.
Every account in Riverside's name. Iris Bites holds zero credentials. They could fire us tomorrow and the system keeps running.
Synthflow
Voice agent infrastructure
$79/mo (their Growth plan)
Twilio
Phone number + call routing
$1.15/mo + ~$15/mo usage
ElevenLabs
Voice synthesis (Adriana voice)
$22/mo (Creator tier)
Anthropic Claude Sonnet 4.6
The 'brain' that reasons about each call
~$28/mo at observed volume
Google Workspace
Calendar + KB Google Doc (they already had this)
$6/user/mo (existing)
Airtable
Captured-leads dashboard
Free tier
Iris Bites retainer
Ongoing tuning, monthly Loom report, support inbox
$97/mo
Total monthly
~$240/mo total
Plus $997 one-time install. Less than a quarter of what a part-time evening receptionist would cost.
After (weeks 2–4 post-launch)
The numbers.
Missed calls / week (before)
~14
Missed calls / week (after Iris)
0 (every after-hours call captured; in-hours overflow caught after 3 rings)
After-hours bookings/week (before)
0
After-hours bookings/week (after, observed weeks 2–4)
3–6
Average AI call duration
1m 47s
Caller-asks-if-AI rate
~18% (Iris discloses honestly when asked)
Caller-hangs-up-on-disclosure rate
~3%
Estimated new-patient revenue captured / month
~$3,400–$6,800 (4–8 new bookings × $850 average new-patient LTV)
From the owner (illustrative)
“We weren't losing patients because we're bad dentists. We were losing them because nobody answered the phone at 6pm. Iris doesn't replace my front desk — Erica still runs the show during the day. Iris catches the calls Erica can't. The math made sense in week two.”
— Dr. Reeves, Riverside Dental of Austin · illustrative composite quote
Want this for your dental practice?
We have an Industry Pilot template ($1,397 + $97/mo) pre-built for dental — same 7-day install, KB starts ~80% configured. Or hear what Iris sounds like first.
Composite case study based on the Riverside Dental KB we use across the leaderboard. Real-customer case studies will replace this once they're live.