Healthcare
A patient asks where their result is. Answering that properly means reaching into the LIS, checking the order is billed and released, and speaking back in the patient's own language — in seconds, at three in the morning.
A chatbot can tell a patient the branch opens at eight. It cannot tell them their biopsy is ready, confirm who may collect it, and book the follow-up.
One patient request, end to end
This is a single WhatsApp message travelling the whole platform. Every step names the module doing the work — the AI is one of nine, not the product.
The message arrives
"نتيجة تحليل الأنسجة جاهزة؟" — Arabic, no reference number, 02:40. It lands in the same inbox as voice, Messenger and the patient portal, on the patient's existing timeline.
Interaction ManagementIdentity and context are built from your systems
Our FHIR agent runs inside your network and is reached over an outbound Redis channel — your LIS, HIS and PACS are never exposed to the internet. It resolves the patient, their open orders and their branch without anyone typing a file number.
Integrations · Robo-FHIRThe AI validates before it speaks
Histopathology is not a same-day test. The AI checks the order status, whether the report is released and whether the balance is settled — and answers from the approved knowledge base, so turnaround times and collection rules come from your policy, not from a model's memory.
AI & Automation · Knowledge BaseIt acts, and books the next step
Result released and settled: the AI sends it through the secure channel, then offers the consultation the report recommends and writes the reservation straight into your scheduling system — slot held, confirmation sent, reminder queued.
AI Agents · CampaignsAnything unresolved becomes tracked work
Balance unpaid, sample rejected, report delayed past its promised date — a ticket opens against the right department with the order attached, an SLA and automatic follow-up if it goes quiet. The patient can ask its status at any time and get a real answer.
TicketingA human takes over with everything in hand
Clinical or sensitive cases route to a qualified agent by skill and language, never to a queue. The co-pilot has the order history and the policy article open, and drafts the reply in the patient's dialect.
Routing · Co-PilotQuality is checked on every case, not a sample
The interaction is scored automatically on close against your own clinical-communication scorecard. Reviewers confirm or override, and agents can dispute — which matters when conduct is regulated.
Quality ControlThe operation learns from it
Result-delay contacts by branch and test type, the questions the knowledge base could not answer, and the staffing the next night shift needs — all fed back before the pattern becomes a complaint.
Analytics · WorkforceWhat we connect to
| FHIR | Patients, orders, appointments, diagnostic reports and coverage through a standards-based interface rather than a bespoke export. |
|---|---|
| Deployed inside your network | The agent sits on your intranet and reaches out over Redis. No inbound firewall rule, no clinical system published to the internet. |
| LIS, HIS, PACS and RIS | Order status, result release, imaging availability and branch scheduling. |
| Scheduling and billing | Reservations written back, balances checked before a result is released. |
Where it is running
Alfa Group — Labs, Scan and Medical Group — has run patient communication on RoboDesk for five years, across a test catalogue that runs from routine panels to NIPT, whole exome sequencing and histopathology. Yodawy handles high-volume pharmacy fulfilment through deep integration with its own platform. Rotana Scan-Lab, Badran Hospital and Purex Health Care run on the same platform, and Egypt's Ministry of Health and Population runs citizen services on it.
See it run on your own traffic.
Thirty minutes with a CX engineer. No slideware.