Home/Resources/Case Studies/Veda — CTCA's AI Assistant
Healthcare & Wellness
Lausanne, Switzerland

Veda — CTCA's AI Assistant

Intercom Fin deployed as a French-speaking AI assistant for a Swiss therapy practice

Veda — CTCA's AI Assistant screenshot

Project Overview

What We Worked On

  • Intercom Fin Setup & Tuning
  • Help Center Creation (29 articles)
  • Website Content Ingestion (251 pages)
  • AI Persona & Tone of Voice ("Veda")
  • Guardrails for Medical Questions
  • Handover Rules to the Practitioner

Background

After rebuilding CTCA's website, Diego Costier was still answering the same questions over and over — which therapy suits which problem, how much a session costs, whether insurance covers it, and how to book. A solo practitioner can't sit in an inbox all day. We deployed Intercom Fin as a named assistant, "Veda", trained on a purpose-built French help center and the whole public site, so visitors get an answer immediately and Diego only gets the conversations that actually need him.

The Challenge

  • Repetitive questions — Therapies, pricing, availability and insurance came up in almost every message.
  • No knowledge base — Everything lived in Diego's head or was scattered across the website; there was nothing an AI could learn from.
  • Health-sensitive context — An assistant in a therapy practice must never give medical advice, assess contraindications, or promise insurance reimbursement.
  • French-first audience — Patients in Lausanne write in French, with local vocabulary around Ayurveda, NAET and auriculotherapy.
  • Trust — Patients need to know they're talking to an assistant, and to reach a human the moment it matters.

The Solution

Phase 1: Knowledge Foundation

Built a French help center from scratch — 29 articles covering therapies, pricing, first-visit expectations, cancellation rules and practical logistics — and connected the 251 public pages of ctca.ch as an additional source, so Fin answers from real, owned content instead of guessing.

Phase 2: Naming & Persona

Named the assistant Veda — Sanskrit for "knowledge" and the root of Ayurveda — and gave it a warm, calm tone matching the practice. Published a public page explaining who Veda is, what it can do, and what it deliberately won't do.

Who is Veda?

Phase 3: Guardrails & Handover

Explicit boundaries: no medical advice, no contraindication assessment, no insurance confirmation. Anything personal or sensitive is routed to Diego directly, with the conversation context attached.

Phase 4: Conversion Paths

Veda doesn't just answer — it guides. Every relevant answer ends on the right next step: the matching service page, the relevant help article, or the online booking flow.

Phase 5: Tuning on Real Conversations

Reviewed real conversations weekly, turned every gap into a new article or snippet, and refined answers until the resolution rate stabilised at 87.5% with no human escalation needed on the vast majority of chats.

Results

Resolution rate

Answered by AI

0% (email only)87.5%

Automation rate

No human touch

0%87.5%

Involvement rate

Fin engages every chat

n/a100%

Knowledge sources

Owned content

None280 (29 articles + 251 pages)

  • A public help center that also works as SEO content and as a self-service layer for patients
  • Questions answered instantly, in French, outside consultation hours
  • Clear boundaries: Veda guides, Diego treats — no medical advice from the assistant
  • Every conversation ends on a next step: a service page, an article, or a booking
  • Low volume, high quality: the point isn't deflection at scale, it's a solo practitioner getting his evenings back

Project Screenshots

Veda, CTCA's AI assistant, answering a question in French about hay fever treatments on the ctca.ch homepage

Veda answering a patient question in French, with the matching help center article and price

Intercom Fin AI Agent performance dashboard for CTCA showing 100% deflection rate and 87.5% resolution rate

Fin AI Agent performance for Veda: 100% deflection rate, 87.5% resolution rate

Hover over cards for 3D effect • Click to pop • Use button to scatter/organize

Technology Stack

Intercom
Intercom Fin
Intercom Help Center
Lovable logoLovable
Supabase logoSupabase

What's Next

  • English and German versions of Veda alongside the site translations
  • Proactive messages on pricing and booking pages
  • Deeper booking integration so Veda can propose live availability

More Case Studies

Ready to Build Your Next Project?

Let's discuss how we can bring your vision to life with the same level of care and expertise.

Start Your Project