Selected Work

Guest messages arrive in several languages across email and booking platforms. Front-desk staff translate by hand, so urgent requests — a leak, a lockout, a medical need — sit behind routine ones, and replies come out inconsistent depending on who is on shift.

The problem was triage as much as translation: knowing what is urgent, answering in the guest's language, and keeping one consistent voice.

An agent that reads every incoming guest message, detects its language and urgency, drafts a reply in the guest's own language in the group's tone, and routes urgent cases straight to staff. Everything else waits for a quick human approval before sending — staff approve instead of translate.

Each store reordered from gut feel and spreadsheets. Fast-moving lines ran out while slow lines piled up, and pharmacists spent mornings on supplier calls instead of patients.

The fix was not just forecasting — it was turning velocity and lead times into a draft order a pharmacist could approve in minutes.

An assistant that watches sales velocity and supplier lead times per store, drafts purchase orders, and flags lines at risk of running out — with every order left for pharmacist approval before it goes anywhere.

Cancellations left empty chairs. Front-desk staff called down waitlists by hand, reminders went out late so no-shows stayed high, and rescheduling meant phone tag.

The work was the loop, not just the reminder: refill every freed slot from the waitlist automatically, and let patients reschedule without calling.

A system that sends timely reminders with one-tap reschedule, and the moment a slot opens, offers it to waitlisted patients in order until it fills. Staff only handle the exceptions.