Sean O'Sullivan

I build AI voice agents and automation that real businesses run on. The fastest way to evaluate my work is to talk to it.

Live demo — answers 24/7
(702) 983-8434

How to test it: pretend you're a patient calling Dental Dynasty, a fictional dental office. Try booking an appointment, asking about pricing, or calling after hours. It's a sandboxed demo of the production system — same AI, no real patient data behind it.

250+calls/week handled in production
245 → 0missed calls for a dental client
<2 minlead response, down from 18
$7K/mobilled automatically at Splash Bros
40 hrsadmin time saved monthly

Production work

Systems I co-built at CallAura, an AI receptionist platform live with paying business clients handling 250+ calls per week. Client names withheld; metrics are from production.

Dental practice — missed calls to zero

AI receptionist deployment, 2025

Problem
Front desk couldn't keep up with call volume. Call logs showed 245 missed calls over the three months before deployment — each one a patient who might not call back.
Deployed
An AI voice agent that answers every call: appointment requests, business-hours logic, live transfer to staff when a human is needed, and after-hours coverage.
Results
245 → 0missed calls after deployment
24/7coverage, including after-hours

Aesthetics clinic — 18-minute lead response to under 2

Automated lead engagement, 2025

Problem
New leads waited an average of 18 minutes for a first response, and 31% of leads were never followed up with at all. In aesthetics, leads book with whoever answers first.
Deployed
Automated AI engagement workflows that respond to every inbound lead immediately, qualify them, and hand warm leads to staff with full context.
Results
<2 minlead response, down from 18
31% → 4%missed lead follow-ups

How it's built

High level only — the interesting parts are in the details, and those stay ours.

Each deployment is a multi-branch voice agent wired into the client's actual operations: quoting, scheduling, transfers, and in-call SMS (including a photo-based quoting flow that prices jobs from customer photos with a vision model). Failure paths are designed, not discovered.

Voice & LLMRetell AI + Claude
Orchestrationn8n workflow automation, webhooks, REST integrations
Business systemsCRM, scheduling, messaging, Stripe payments
My roleCo-founder & engineer — architecture, integrations, prompt design, testing, live production support

Splash Bros — internal field-service platform

Sole engineer for my brother's pool service company in Las Vegas. Built the customer site and the internal platform the business runs on: 50–60 recurring accounts serviced weekly, in production since 2024.

Problem
Every service visit generated paperwork: chemical logs, photos, billing, customer notifications. It ate roughly 40 administrative hours a month.
Built
A mobile workflow technicians complete on-site — chemical tracking, service documentation with photos, billing, and customer notifications fire automatically. Stripe billing handles ~$7K/month in payments.
Results
~40 hrsadmin time saved per month
11 minsaved per service visit
2 yrsin production, still running

This was a v1 built solo, with no budget, for exactly one customer: the business itself. It isn't pretty — it's adopted, which is harder.

One tap runs the back office

The technician fills out one form poolside and hits Submit. Everything below happens automatically — no office work, ever.

Service form: before photo of an algae-green pool, technician notes 'Pool has heavy algae, shock treatment done', chemical amounts entered Service form continued: after photo of the same pool, now clean and blue, ready to submit
1 · Technician fills one form poolside — before/after photos, notes, chemicals — and hits Submit. (Real service: heavy algae to clean in one visit.)
Automatic customer email: service timestamp, technician notes, itemized chemicals used, before and after photos attached
The customer instantly gets a service report — notes, itemized chemicals, both photos attached. Same submission, same timestamp.
Stripe upcoming invoice: chemicals metered per unit (tablets, gallons, pounds) rolling up onto the monthly bill alongside the base service
Chemicals are metered per unit through Stripe usage-based billing and roll onto the customer's monthly invoice automatically — a month of visits, aggregated, zero manual entry.
Google Sheets usage log: 200+ timestamped service rows with chemical amounts and billing status, customer columns redacted; test submission highlighted
Every visit is logged to a running service history — 200+ rows and counting, each marked when billed. The highlighted row is the submission shown above, logged seconds after Submit. (Customer details redacted.)