Cut 240 hours/month of manual patient operations.
- Hours/month saved
- 240h
- Avg response time
- 12s
- Supported languages
- 4
- Triage precision
- 94%
Context
A clinic with 8 specialties, 4 patient languages and a 6-person reception team drowning in repetitive work.
Challenge
Reception spent 70% of its time on mechanical tasks: appointment confirmation, answering the same 30 FAQs, translating documents, triaging by urgency. Triage errors meant real emergencies waited while routine consults burned specialist slots.
Solution
We designed and deployed an AI agent connected to the HIS (clinical records) that: (1) answers FAQs in 4 languages via WhatsApp and web, (2) does initial triage following clinical protocols signed by medical leadership, (3) suggests and confirms appointment slots, (4) escalates to humans on red flags. Reception moved to handling exceptions and complex cases.
KDK Intelligence started with a 2-week Express Diagnosis. We identified 11 opportunities; the 3 prioritised drove 80% of the return. We moved to a Standard Development sprint, delivered in 75 days, and the clinic hired Standard Support at close.
Compliance was critical: everything was built against GDPR and Spain’s specific healthcare data protection rules. Patient data never leaves the HIS. The agent works on a redaction layer that masks identifiers before they reach the LLM.
Internal adoption worked because we designed the system with the reception team, not for them. All 6 people still work there — and now they work on what actually matters.
"We went from putting out fires all day to actually improving patient experience. Reception stopped being the bottleneck."
