Multichannel intake
Web, WhatsApp and voice normalise into one input contract before any processing occurs.
WTTT Systems · Operational Infrastructure for Healthcare
Conversational patient intake, automated triage, and conflict-proof calendar synchronisation through language models with strict validation.
./product
An agentic layer that handles the request in conversation, decides its urgency, and writes to the calendar as a single atomic operation. The model proposes; the system validates, locks, and records.
Web, WhatsApp and voice normalise into one input contract before any processing occurs.
Intent and urgency are classified before touching the calendar, separating routine requests from cases that need a person.
Proposal, confirmation and calendar write as one idempotent operation, with conflict resolution.
./architecture
Design reference for the platform in Private Alpha.
Web, WhatsApp and voice normalise into a single contract before any processing.
Classifies the request and decides which model handles it and whether it must escalate to a person.
Sonnet for multi-step reasoning and disambiguation; Haiku for low-latency triage.
Strict JSON schemas for reading availability, booking and confirming. No free-form writes to the calendar.
Emergencies, clinical ambiguity and scheduling exceptions escalate to a person with full context.
./security
The same request processed twice never produces two appointments. A booking is confirmed by idempotency key before the calendar is touched.
Every model output is validated against a JSON schema before a tool executes. An invalid schema executes nothing.
Clinical data never travels inside the prompt. The agent operates on references, and access to the data stays inside the clinic's own system.
Every agent decision is recorded: what was requested, which tools ran, and what result they produced.
Urgent symptoms, clinical ambiguity, and scheduling conflicts are never resolved automatically.
./company
WTTT Systems is a software company building agentic infrastructure for the healthcare sector. We build the layer that lets a clinic operate autonomously without delegating clinical judgement to a model.
We work on systems where a mistake is not corrected with a rollback: it is corrected with a person. That is why the architecture prioritises determinism, traceability and human supervision over maximum automation.
./contact
Onboarding capacity is limited. Tell us your clinic's volume and which part of the flow you want to automate first.
This link opens your email application; no data is submitted through this site.