"How much do facial fillers cost?" is one of the most common questions aesthetic clinics receive — on Instagram DMs, Messenger, Viber, phone, and the website contact form. The question sounds simple, but the honest answer is not. Price depends on the product brand, the volume used, the treatment area, the practitioner's experience, and whether it is a first session or a touch-up. That mismatch — a simple question, a nuanced answer — is where clinics lose hours every week and where prospective clients quietly drop off.
At Pragma AI we work with clinics that want to keep the human consultation sacred, but stop drowning in repetitive first-touch messages. Here is what the automation actually does, step by step, so you can judge whether it fits your practice.
Step 1: One inbox, many channels. Enquiries arrive on Instagram, Facebook, Viber, WhatsApp, the website form, and sometimes email. We connect these channels into a single conversation view using n8n self-hosted on your infrastructure (or Hetzner, under your account). No message is left in a silo. Your front-desk team stops toggling between five apps.
Step 2: A first-touch assistant, not a pricing oracle. The AI assistant does not quote a final price. It does what a well-trained receptionist would do: greets the person in Bulgarian or English, acknowledges the enquiry, explains that filler pricing depends on the specific product and volume discussed during consultation, and offers the next concrete step — booking a consultation slot, sending your published price range document, or routing to a human when the question is medical.
Step 3: Structured intake. Before a human ever reads the thread, the assistant collects the useful context: which area the person is asking about (lips, cheeks, nasolabial folds), whether they have had fillers before, and their preferred contact channel and time. This turns a vague "колко струва?" into a briefed enquiry that your practitioner can act on in two minutes instead of ten.
Step 4: Calendar and reminders. Once a consultation is booked, the workflow sends a confirmation, a pre-consultation form (allergies, medications, previous treatments), and a reminder the day before. Fewer no-shows, less admin, and the client feels looked after before they walk in.
Step 5: You stay in control of the medical conversation. The assistant is explicitly instructed to hand off to a human the moment the conversation touches contraindications, medical history, or anything resembling clinical advice. Filler treatment is a medical act; the automation handles logistics around it, never the act itself.
Where the data lives. For aesthetic clinics we default to self-hosted infrastructure — your Hetzner or AWS account, your database, your backups. Conversations, phone numbers, and intake forms do not sit in a third-party marketing tool you cannot audit. That matters when you are handling health-adjacent information.
What we do not do. We do not publish price lists you have not approved. We do not promise treatment outcomes. We do not send marketing broadcasts on Viber or Messenger without opt-in. We do not replace your practitioner in the consultation. The assistant is a front-desk teammate, not a clinician and not a salesperson.
What a realistic setup looks like. A typical clinic starts with the Instagram DM channel plus the website form, because that is where most "цени" questions land. We add Viber and phone routing in a second pass, once the first flow is calm and your team trusts it. The whole engagement is built and run by our 4-engineer team — we are accountable for the system working, not just for delivering a diagram.
If your front desk is spending its mornings answering the same three questions about filler pricing before anyone has been examined, that is exactly the workload this is designed to lift.
Talk to us. If you run an aesthetic clinic in Bulgaria and want to see what a first-touch assistant would look like on your own channels, contact Pragma AI at pragma-ai.eu. We will walk you through a specific setup for your practice, using your real enquiry patterns — not a generic demo.