Back to case studies
IN PRODUCTIONGeneral & aesthetic medicine · Barcelona

Sermed Clinic — 28% of Reception Capacity Released Through AI and Process Redesign

Sermed Clinic's reception team supported roughly 2.5 FTE of capacity against 80–100 weekly inquiries. Aigenrix redesigned the workflow so AI now handles the predictable share of first contact and booking — releasing an estimated 28% of that capacity, worth roughly €19,300 a year, for an estimated 176% first-year return.

Sermed Clinic — 28% of Reception Capacity Released Through AI and Process Redesign
< 2 min
First response, 24/7
~80%
Inquiries handled by AI at first contact
~28 hrs/week
Reception capacity released
~28%
Of original reception capacity released

The challenge

Sermed Clinic is a five-specialist general & aesthetic medicine practice in Barcelona. Inbound demand averaged 80–100 appointment inquiries per week, arriving across WhatsApp, Instagram, web chat and direct calls. Roughly 2.5 full-time equivalents on the reception side covered communication, patient qualification, scheduling, confirmations and follow-up — a real cost the clinic had never quantified as a single number.

A significant share of this work consisted of predictable, repetitive administrative tasks. At the same time, inquiries arriving in the evening, overnight or over the weekend could sit unanswered until staff became available. The core constraint was structural: as inquiry volume grew, administrative workload grew with it, while response speed stayed dependent on who was on shift.

What we found

Repetitive first-contact work

Reception staff manually handled standard questions, initial patient qualification and routine appointment requests. Human capacity was being used for predictable administrative work that did not require clinical or business judgment.

An availability gap

Inbound demand could only be processed immediately when reception staff were on shift. Warm patient demand routinely sat waiting outside normal working hours — evenings, nights and weekends.

A manual booking workflow

Qualification, scheduling, confirmations, reminders and parts of cancellation handling all required repeated manual intervention. Administrative workload scaled almost directly with inquiry volume.

The problem was not the absence of another software tool. The underlying workflow itself needed to change.

What we built

Aigenrix redesigned the intake and booking workflow so AI handles the stages that don't require judgment, and staff stay in control of exceptions and anything non-standard. An AI agent now owns first contact across every channel — answering FAQs, qualifying patients against the clinic's criteria, and booking standard cases straight into the calendar.

Confirmations, reminders and structured patient notes are generated automatically, and a no-show layer — prepayment on at-risk slots, automated attendance confirmation, proactive calendar control — catches missed appointments instead of letting them shrink utilization silently. Complex cases and every clinical decision stay with the doctors; the agent escalates rather than guesses.

Before

Inquiry → Wait for receptionist → Manual response → Manual qualification → Scheduling coordination → Booking → Confirmation → Follow-up

After

Inquiry → Immediate AI response → Automated FAQ handling → Structured qualification → Automated booking for standard cases → Confirmations and reminders → Human escalation where required

Results

  • First response is now under 2 minutes, 24/7 — the clinic is no longer dependent on reception working hours for initial digital contact.
  • Approximately 80% of inbound inquiries are handled entirely by AI at first contact, concentrating staff attention on exceptions, complex situations and higher-value interactions.
  • Observed capacity release ranges from roughly 25 to 35 hours per week; the base case used for the figures on this page is 28 hours per week.
  • 28 hours per week is equivalent to roughly 0.7 FTE of reception capacity — about 28% of the original 2.5 FTE reception function.
  • Better slot utilization through prepayment workflows and automated attendance confirmation.

Economic impact

~€19.3K
annualized capacity value
~176%
estimated first-year capacity-value ROI
~4.3 months
estimated payback
Receptionist gross salary benchmark€2,300/mo
Annual gross salary€27,600/yr
Annual working hours2,080 hrs
Gross salary equivalent€13.27/hr
Released capacity, annualized1,456 hrs/yr
Implementation investment€7,000

The ~€19.3K figure is the annualized gross-salary-equivalent value of released capacity — not automatically realized cash savings, payroll reduction or profit.

That value is realized when the clinic uses the released capacity to avoid additional hiring, serve more patients, respond faster, improve calendar utilization or reallocate staff to higher-value work.

Business transformation

Before

More inquiries → More manual work → More administrative load → More headcount pressure

After

More inquiries → AI handles predictable operations → People handle exceptions → Existing team gains additional capacity

The primary outcome was not the introduction of AI. It was a change in the economics of the reception process.

Key insight

Sermed Clinic did not need AI for the sake of AI. A significant share of human capacity was being consumed by work that did not require human judgement. Aigenrix identified the operational leakage, redesigned the workflow, introduced AI where it was economically justified, and measured the result: roughly 28% of reception capacity released, worth an estimated €19,300 a year in capacity value, for an estimated 176% first-year return and a payback of about 4.3 months.

What stayed human

  • High-value consultations remained fully doctor-led from first contact through follow-up.
  • Final treatment decisions stayed manual — the AI never touches clinical judgement.
  • Complex patient scenarios continued to route to staff for human handling.

Honest tradeoffs

Where the rollout was not instant.

  • The first weeks required active manual supervision of AI replies before autonomy increased.
  • Tone calibration — matching the clinic's voice and Spanish medical phrasing — took roughly 2–3 weeks.
  • Edge-case patient scenarios still need staff intervention; the agent escalates rather than guesses.
  • Some conversational flows needed iterative optimization across the first month before they read naturally.

Built on

AI brain — Claude (Anthropic)Voice & STT — DeepgramInfrastructure — AWSCompliance — GDPR readyPayments — StripeMessaging — WhatsApp & Web chat

Want to find the equivalent leakage in your business?

30 minutes. We will assess your situation and tell you honestly if we can help.