We built an AI agent hub for aesthetics clinics: behind the scenes

This is a behind the scenes piece. We share why our agency in Brazil decided to build an AI agent hub for aesthetics clinics, which agents live in it today, what we learned along the way and what we decided not to do. No made up metrics, no promise of guaranteed results, just what we saw running with paying clients.

Why we built the hub

Most aesthetics clinics that reach us have the same bottleneck: sales. Good ads bring leads, the front desk cannot keep up, the lead goes cold, the clinic blames traffic. After repeating this diagnosis dozens of times, it became clear that optimizing campaigns was not enough. The missing piece was what happens after the click.

Applied AI, especially Claude connected via MCP to the clinic's systems, became the natural answer. Instead of selling only traffic, we started implementing agents that live inside the sales process.

The agents that live in the hub today

1. Support and triage agent

Receives the WhatsApp message, understands which procedure the lead is asking about, runs basic qualification questions and hands over to the human front desk only when the lead is ready to book. On late nights and weekends, it responds instead of nobody.

2. Content agent

Drafts Reels scripts, captions, procedure descriptions without forbidden claims (respecting Brazilian aesthetics council rules) and post ideas for seasonal moments. The clinic edits and publishes; the agent does not post on its own.

3. Quote agent

Based on the procedure of interest and the qualification answers, it assembles the quote in the clinic's standard format, including conditions, payment terms and validity. The front desk only reviews and sends.

4. Follow up agent

Every day it prepares a queue of messages for leads who did not reply to the quote, patients who did not come back and birthdays of the month. The front desk approves the queue; the actual send is human.

5. Reporting agent

At month end it reads campaign, calendar and CRM data. It assembles the clinic report in our agency's consultative format, with what worked, what to fix and a per procedure read.

What we learned building the hub

1. Without a good process, AI accelerates chaos. Before plugging an agent into a messy clinic, we fix the sales flow on paper.
2. Regulation comes first. Aesthetics clinics in Brazil have specific rules from CFBM and CFM. Every agent lives inside those limits, no exception.
3. Humans decide, AI prepares. Closing a sale, clinical decisions and sensitive cases stay with the front desk and the professional. The agent sets the stage.
4. Text context rules. The better the clinic's material (procedures described, conditions, refund policy), the better the agent responds.
5. Starting small is the only path that lasts. A clinic that tries to turn on five agents at once gives up in two weeks.

What we decided not to do

• Promising a specific number of bookings or conversion rate. We do not know the outcome for your clinic, only the general process history.
• Closing sales without a human. The front desk is irreplaceable at the moment of yes.
• Automating before standardizing. If every attendant answers differently, the agent learns chaos.
• Selling a standalone agent. The hub makes sense inside the MADS method, together with paid traffic.

What changes in a clinic once the hub is running

How we use this in the agency's daily work

The 571+ companies our agency has accelerated went through versions of this flow. In the clinics where the hub runs end to end, the human consultant spends less time firefighting operations and more time thinking strategy. That is what the agency delivers: sales predictability so the clinic only decides what actually requires human judgment.

Where to keep reading

Details on each agent and how they connect to the sales process: AI agents for aesthetics clinics. Other agency cases (diagnosis quiz, reports, SEO+GEO on this very site): real use cases. To start: free diagnosis.