Inbound patient contact routing: patient chat and calls received on the provider's channels, classified by clinical or administrative role, with escalation to the on-duty physician for cases that require clinical attention.
An AI agent squad for the healthcare and life sciences industry.
Polyant is the framework for building AI agent squads for healthcare providers: routing of inbound patient contact with physician escalation, draft of the medical record after the visit, optimisation of operating rooms and beds, audit of clinical systems and electronic health record. Consistent with AI Act, MDR, GDPR, national healthcare authority constraints.
Polyant puts an AI agent squad to work on the operational flows of the healthcare provider:
- post-visit medical record dictation with ICD-10 coding, prescriptions and follow-up; patient intake arriving from different channels (phone to reception, booking portal, walk-in at the front office, chat or app where they exist) with recognition of situations that require immediate clinical attention
- audit of clinical systems (HIS, LIS, RIS, EHR) for privacy compliance and access traceability; planning of operating rooms and beds under operational and clinical constraints
- the runtime audit registry stays queryable by the DPO and the head of compliance via standard SQL on the provider's database. The MDR-SaMD avoidance framing is cardinal: the agents do not diagnose, do not prescribe, do not replace clinical judgement
Four typical contexts of the healthcare provider.
The areas where the agent squad supports the clinical and operational team.
Clinical documentation: ambient transcription of the medical visit with ICD-10 coding and draft of the electronic medical record for the physician to approve.
Operational planning: optimisation of operating rooms with LOS forecasting, bed management, patient allocation.
Clinical system audit and compliance: periodic audit of accesses to electronic health record systems and clinical systems (HIS, LIS, RIS) with anomaly identification, national healthcare authority and GDPR compliance.
Five coordinated agents for the clinical-administrative flow.
The patient cycle in a healthcare provider is not a single step. It is an end-to-end flow across reception, outpatient clinic, reporting and audit. Five coordinated agents that do their part, under the same configuration and the same audit registry.
Patient Acquisition Triage
Routing of inbound patient contact (chat, phone to reception, booking portal) classified by service and preliminary urgency, with escalation to the on-duty physician for cases that require clinical attention.
See the agentPatient Communication Reviewer
Verification of patient communications pre-send (newsletters, reminders, clinical communication) for health data protection.
See the agentPatient Communication Reviewer
Pre-send verification of patient communications (newsletters, visit reminders, clinical communication) against health data protection and the medical-legal registry.
See the agentClinical System Audit
Periodic audit of accesses to electronic health record systems and clinical systems (HIS, LIS, RIS) with anomaly identification, national healthcare authority and GDPR compliance.
See the agentAudit Recorder
Records every step of the squad in the audit registry: EHR access, routing decision, clinical classification. The DPO queries via standard SQL for national healthcare authority audit or data protection authority inspection.
See the agentNeed other agents for the healthcare provider? We build them together.
Five coordinated agents for the clinical-administrative flow are an example. Every provider has different systems (proprietary HIS, regional EHR, clinical management software), specific regulations (national healthcare authority, region, professional board), different internal procedures. The agents that truly serve get built inside the project.
Where the agents meet healthcare regulations.
Healthcare regulation is wide and evolving. Polyant does not promise full compliance: it states what the product does day to day to support the requirements the customer's team has to meet.
AI Act: MDR-SaMD avoidance framing
The agents do not diagnose, do not prescribe, do not replace clinical judgement. The MDR-SaMD avoidance validation verifies that new agents stay out of the medical device software perimeter. Agents: MDR-SaMD Avoidance Checker, AI Act High-Risk Validator.
MDR, Medical Device Regulation
The declarative rules written by the clinical team define what the agent can do and what it cannot do. Clinical decisions stay with the physician. Agent: MDR-SaMD Avoidance Checker.
GDPR and health data (art. 9)
Health data is a special category. The runtime audit registry traces every access to patient data, inspectable by the DPO. Agents: Audit Recorder, Policy Mapper.
Electronic health record and access traceability
Audit of accesses to the electronic health record with periodic reporting to the head of healthcare authority compliance. Agent: EHR Audit.
National healthcare authority guidelines for clinical systems
Verification of HIS, LIS, RIS conformity to the regulation and traceability of automatic clinical decisions. Agent: Clinical System Audit.
NIS2 for healthcare providers as essential service operators
The public healthcare provider falls within the NIS2 perimeter: incident notification to the CSIRT within the regulatory windows. Agent: NIS2 Incident Reporter.
Open transparency for the shared responsibility and AGPL detail →
Four questions from healthcare heads.
No. The MDR-SaMD avoidance framing is cardinal. The agents support the physician (pre-booking triage, document summarisation, recognition of critical signals for human escalation, monitoring of chronic data), they do not replace clinical judgement. Diagnosis and prescription stay with the physician.
For every patient interaction involving health data, consent is a prerequisite. The declaration that the interlocutor is an automated system is mandatory for patient-facing systems. Consent management rules live in the customer's repository, versioned.
Electronic medical records and EHR systems are heterogeneous per provider and region. The integration is delivered during the project by the Exelab team through the APIs of the specific system.
The typical pattern is 14-22 weeks (longer than other industries because of the depth of clinical triage rules and the validation of the MDR-SaMD perimeter with the compliance team).
From a 30-minute conversation to the squad in production.
A 30-45 minute conversation to understand how Polyant for healthcare applies to the provider's case and how long it takes to bring the clinical-administrative squad into production.