One owner, a facility open around the clock, and three automation layers handling the calls, texts, and emails that used to land on his phone.
The Birdie Collective is a 24/7 indoor golf simulator lounge in St. Charles, Illinois. Members badge in and play at two in the afternoon or two in the morning. For most of those hours there is nobody on site.
Derek Cooper is the owner and the day-to-day operator. Every inbound call, every text, every membership signup, every punch-pack purchase, and every cancellation reached him personally, at whatever hour it happened. As the lounge grew, that stopped being workable.
A round-the-clock facility takes round-the-clock questions. Are bays available? How do I book? Can we bring in alcohol? All of it landed on one phone.
New memberships, punch packs, and league registrations each needed a welcome message with the right access instructions. Sent by hand, or not at all.
In Derek's words, as the business got busier “it started getting more and more to handle that all myself.”
This was not a single deployment. It started with the phone: an AI voice agent trained on the lounge's booking rules, membership tiers, hours, and policies. It answers naturally, handles interruptions, and hands off to Derek when a caller genuinely needs a human. Since April 2026 it has been scoped to cover the hours when nobody is on site.
Next came booking SMS. A Square webhook pipeline watches every payment and order event, so a first-time booker receives a welcome text with access and how-to-play instructions, followed by a reminder before their session. Then came the membership email lifecycle, which routes signups by type so each membership tier, punch pack, and league registration gets its own onboarding, and a cancellation triggers a feedback request instead of silence.
Each layer was added only once the previous one was running cleanly. That is why the system now covers the whole customer lifecycle rather than one slice of it.
The entire chain runs unattended. A customer transaction is the only trigger required.
Square and the booking platform emit an event the moment someone pays, books, joins, or cancels. Every event is logged and de-duplicated so nobody is ever messaged twice.
➞The system checks whether this customer has been seen before, what they bought, and which journey applies. A first-time booker and a renewing member get different messages.
➞The right text or email goes out on schedule, with the customer's name and the correct access instructions. Delivery is written back to the log so every send is verifiable.
Every figure below comes from the system's own production logs, not an estimate.
A membership welcome and a punch-pack receipt are not the same conversation, and a cancellation is a chance to learn something rather than a dead end. Four journeys stack to the 102:
Voice agent live since November 2025, narrowed to unstaffed hours in April 2026. Payment and email automation live since February 2026, booking SMS since April 2026. All counts current as of August 2026 and taken directly from production logs.
“I didn't want something that was robotic. I wanted something that felt very real, very conversational, very human. And this by far was the most realistic voice agent that I ever experienced.”
“If I ever need a change or an edit, I can email or call Michael up and he's usually on it within that same day. I don't have to call a 1-800 number to get an answer. I could actually call the owner of the business and get an answer, which is absolutely huge.”
Generic autoresponders fire on a schedule and hope. This runs on what the customer actually did, and it proves each send.
Messages fire off real payment and booking events, so timing matches what the customer just did rather than an arbitrary cadence.
Every event is checked against the processed log first. Across 7,181 events, no customer was messaged twice for the same action.
Four distinct email journeys and two SMS sequences, each carrying the access instructions that actually apply to what was purchased.
The voice agent transfers to the owner when a caller needs a person. It escalates rather than guessing.
Delivery status is written back for each message, which is how a 100% delivery rate across 144 texts can be stated as fact rather than assumed.
Voice first, then SMS, then email lifecycle. Each layer proved itself in production before the next one shipped.
If calls, signups, and follow-ups are landing on one person's phone at every hour, that is a solvable problem. Let's map it out.
Book a Call →