Milo: AI WhatsApp booking agent for sports clubs
How we built and launched an AI agent that books football and padel courts end-to-end inside WhatsApp, live with 10 Argentine clubs in 3 months.
| Field | Value |
|---|---|
| Product | Milo |
| Industry | Sports & Recreation | SMB SaaS |
| Location | Argentina |
| Challenge | Small and mid-sized sports clubs had no booking software they would actually adopt; existing apps pushed friction onto players |
| Result | Launched in 3 months and live with 10 clubs on a 10% commission-per-booking model |
Situation
In Argentina, booking a football 5 pitch or a padel court is almost always a WhatsApp conversation. A player messages the club, asks what times are open on Saturday, negotiates an hour, and eventually sends a deposit screenshot. Multiply that by dozens of players per club, per night, and the problem becomes obvious: WhatsApp is the booking channel, and no one is managing it well.
The software market had not solved this. Booking apps exist, but they force players to install an app, create an account, and learn a new interface before they can make a single reservation. Clubs that had tried them watched their conversion drop. There was no dominant incumbent serving the small-to-mid segment of Argentine sports clubs. What existed was paper notebooks, shared Excel files, and the owner's personal WhatsApp on do-not-disturb mode every weekend.
Challenge
The problem had two faces that most solutions only addressed one of.
For players, friction killed bookings. Any step between "I want to play Saturday at 9" and "confirmed" was a place the reservation could die. App downloads were a hard stop. Sign-up forms were nearly as bad. Every booking tool that assumed the player would adapt to its interface was solving the wrong problem.
For owners, operational overhead had metastasized. A single mid-sized club fielded dozens of inbound WhatsApps every evening, often past midnight. Availability lived in a paper notebook or a spreadsheet that only one person kept up to date. Deposits were requested via bank transfer screenshots that had to be manually reconciled, if they arrived at all. No-shows for unpaid slots were common. There was no real-time view of weekend occupancy, no way to know which hours were underperforming, and no structured history of customers.
The clubs were not asking for software. They were asking for the evening back.
Solution
We built Milo as an AI agent that lives inside the channel clubs and players already use. The product was designed, built, and launched over three months as three coordinated applications: the player-facing WhatsApp agent, a multi-tenant backoffice for club owners, and the marketing and onboarding surface for new clubs.
An LLM agent that actually handles the conversation. The core of Milo is a large language model agent that negotiates bookings in natural Argentine Spanish. It handles partial requests ("algo el sábado a la tarde"), proposes alternatives when the requested slot is taken, answers questions about the club, and keeps context across a conversation. It is not a scripted flow with menu buttons. For players, it feels like messaging the club, because that is what they are doing; the difference is that the club now responds instantly, at 2am, on a Sunday.
Payments collected inside the chat. Once a time is agreed, Milo generates a MercadoPago deposit link and collects it mid-conversation. The booking is only confirmed when the deposit clears. This single change eliminated a class of problem clubs had lived with for years: reservations made, never paid, never honored, and a court empty on a Saturday night.
A backoffice built for the way clubs actually operate. Owners and staff use a multi-tenant web app with a real-time schedule, turn management, blocking periods for maintenance or tournaments, a customer dashboard, and role-based permissions for team members. Every slot Milo books shows up instantly. Everything is one source of truth.
Zero player-side friction, by design. No app install. No account. No password. A player who has never heard of Milo can complete a booking in the same WhatsApp thread they would have used to message the club anyway. The product meets both sides exactly where they already are.
Results
Milo launched and scaled to 10 paying Argentine clubs in under three months, across football 5, football 7, and padel. The business model, a 10% commission on each booked hour, aligned incentives with the clubs: Milo earns only when the club does.
| Outcome | Impact |
|---|---|
| Time to market | 3 months from first line of code to live product |
| Clubs onboarded | 10 Argentine clubs on a 10% revenue-share model |
| Player-side friction | Zero app installs, zero sign-ups, zero accounts required |
| Owner workload | Owners stopped manually answering nightly WhatsApps |
| No-show risk | Deposits collected upfront via MercadoPago before confirmation |
| Operational tooling | Paper notebooks, Excel sheets, and group chats replaced by a single dashboard |
| Visibility | Real-time occupancy and customer history for owners, live |
The shift for clubs was not a feature upgrade. It was a change in what the evening looks like. The owner is not a dispatcher anymore. The agent handles the conversation, takes the payment, books the slot, and files it in the schedule. The owner runs the club.
Built in 3 months. Live in 10 clubs. Spanish-speaking, WhatsApp-native, and payment-complete on day one. Milo is proof that the right product does not fight the user's existing behavior, it automates the other side of it.
