What Amigo Does

A concrete scenario showing what Amigo does - from phone call to EHR writeback - and why the data foundation matters.

A patient calls their doctor's office at 2:47 PM. Before the first ring finishes, the AI agent is already connected and greets them: "Hi, this is Dr. Chen's office, how can I help you today?" Zero dead air.

The system has already identified the caller from their phone number, pulled their record from the EHR, and loaded their upcoming appointments and recent encounters into context. The patient asks to reschedule their Thursday appointment. The agent searches available slots, confirms the new time, and books it.

Midway through the call, the patient's voice shifts. They sound frustrated. The emotion detection system picks this up from vocal prosody and adjusts: the agent slows its pace, softens its tone, and acknowledges the difficulty. The patient mentions they lost their insurance card. The agent captures the new insurance details from the conversation and writes them to the world model at voice confidence (0.5), where they'll be verified before syncing back to the EHR.

After the call ends, the system generates a clinical summary, verifies the extracted data against the transcript, and queues the confirmed changes for EHR writeback. An operator reviews flagged items in the review queue. The whole interaction took 3 minutes.

This is what Amigo does: it runs the operational layer of healthcare communication so clinical teams can focus on clinical work.

The Data Foundation

The intelligence and reliability of every system on the platform - voice agents, connector runner, operator workflows, outbound campaigns - comes from the data layer underneath them.

Healthcare data is messy. Outside of billing, RCM, and some operational tables, most of it is unreliable. Clinical notes are unstructured free text. Scribe outputs vary by provider. EHR inputs are often copy-pasted templates with stale information carried forward from visit to visit. Inbound data from patients is worse: callers give wrong dates of birth, forget which insurance they have, or sometimes are not who they claim to be. Some calls are junk entirely. The external systems that hold patient records expose wildly different APIs with unpredictable throughput and availability.

Amigo's answer is the world model - an event-sourced data core that absorbs data from every source, scores it by confidence, cleans and verifies it through automated review pipelines, and projects a unified picture of every entity (patients, appointments, providers, locations). Agents, connectors, and analytics all read from the same verified state. When the world model's data is good, everything running on top of it is effective. When incoming data is bad, the confidence gates catch it before it propagates.

This is the core value proposition: a data foundation that turns unreliable healthcare inputs into trusted, unified state - and an infrastructure layer that handles the messy reality of exchanging data with external systems so that agents and integrations can operate on clean, verified information.

But the world model does more than store data - it makes data exchange at unreliable external interfaces work. When the agent books an appointment, the intent is captured as an event in the world model immediately. The EHR gets the write when it can handle it - seconds later, minutes later, or after a retry cycle. The patient gets confirmation from the world model's state, not from the EHR's response. This decouples the patient experience from the reliability of the external system. The patient hears "you're confirmed for Thursday at 10 AM" because the world model says so, not because an EHR API returned 200 OK in time. The EHR sync happens in the background, with the connector runner handling retries, conflicts, and eventual consistency. The result: patients get immediate, reliable service even when the systems behind the scenes are slow, flaky, or temporarily down.

Two API Surfaces

For a detailed comparison of the two APIs, including authentication, data models, and SDK support, see Classic vs Platform.

Classic API - Text-based digital health. Chat agents for consumer apps.

Platform API - Enterprise voice and healthcare operations. Phone calls with EHR integration.