The challenge
The studio had outgrown its patchwork of tools: a third-party booking widget, a separate payment flow, spreadsheets for attendance, a ticketing inbox for rescheduling, a generic CRM that no one trusted, and no meaningful quality assurance on customer conversations. The experience was fragmented for customers and exhausting for staff. Growth was blocked because the operating model could not scale.
The solution
We took on the full digital stack as a single coordinated engagement and delivered it as one product. Five interlocking builds, one platform, one data model, one operating model.
One engagement, one platform
The studio did not need a booking widget or a CRM vendor. It needed someone to treat the customer journey as a single system and build it that way. That is what we did.
1. Reservation system & customer web application
The core of the platform: a single web application covering the full end-to-end customer journey.
- Discovery (classes, therapists, packages, pricing)
- Profile and medical-grade intake for the therapy side
- Reservation and secure payment
- Smart reminders and one-click rescheduling
- Attendance, feedback loop and retention signals
- A unified profile the customer controls
Every touchpoint was designed around the customer, not the back-office tools.
2. Customer-support chatbot
Grounded in the studio’s catalogue, class schedule, therapist profiles and an internal knowledge base. Handles routine questions (opening hours, rescheduling, package rules, cancellation policy), recommends the right class or therapist based on the customer’s history, and hands off to a human when it does not know. It never gives medical advice — any medical question escalates to a human and is logged.
- ~65% of inbound questions resolved without a human
- 24/7 availability, including German, English and Czech
- Median response time under 15 seconds
- Full audit trail of every interaction
3. Voice bot for quality assurance
Before we arrived, QA meant a senior staff member listening to a handful of recorded calls per week. We replaced that with a voice-capable agent that reviews 100% of customer calls (recorded under explicit consent), scores each against the studio’s rubric, writes a short coaching note, and flags any call where a customer sounded unhappy or a service promise was at risk.
- 100% call coverage, versus 2% manual sampling before
- Mean time from call to coaching note: under 6 minutes
- Objective, consistent QA the studio owner can trust
4. Ebook-creation feature for user experience
A differentiator for the premium tier. Customers can generate a personalised ebook — a recovery guide, a training plan, a nutrition companion — assembled from the studio’s licensed content library. The agent is strictly grounded in that library; it does not invent medical guidance and it cites every passage. Therapists review and co-sign any ebook that touches clinical content.
- 3× engagement lift on cohorts that used the feature
- Opened a new premium tier
- 0 clinical safety escalations
5. Custom CRM
A CRM designed around the studio’s real operating model — not a vendor’s. Every customer record is enriched in real time by the agent layer: last interaction, upcoming bookings, open issues, retention signal, preferred therapist, contraindications. Staff open a single screen and everything is already there.
- Three previous tools retired
- Staff time on CRM admin reduced ~70%
- Compliance-ready audit trail for every touchpoint
The operating model underneath
Every component shares one data model, one identity layer, one audit log and one governance baseline. That is the point. A fitness and healthcare studio cannot afford a stack of tools that disagree about the customer. One platform, coordinated by one team.
Privacy, by design
Intake and clinical data are handled as health data under GDPR with explicit consent, strict retention and full audit of access. All processing happens in the EU. The voice-bot QA pipeline is consent- based and pseudonymised.
Services deployed
- AI Strategy & Integration (journey design and operating model)
- Custom AI Agents (chatbot, voice QA agent, ebook generator, CRM enrichment agents)
- Knowledge Engineering (catalogue grounding, intake/clinical retrieval)
- AI Governance (consent, retention, audit, clinical escalation rules)