AI medical characters

Intelligence that knows its place.

People with roles, context and memory. An authoritative simulation decides which actions can change the world.

Local dialogue prototype · perception and speech input planned

Discuss your use case Explore the complexes

01 / Simulation architecture

Role-aware, context-aware.

An admissions colleague and an orthopedic surgeon should see different information. MediVerse separates public conversation, assigned case context and the permissions needed to act.

Persistent identities and conversational memory
Assigned-patient context and professional roles
Knowledge boundaries between staff, patients and visitors

02 / Simulation architecture

Intent passes through a gate.

A language model can propose an action. Identity, authority, case version and simulation rules must validate it before patient state changes.

Propose → validate → execute → record
Auditable clinical actions and state versions
No clinical changes from dialogue alone

03 / Simulation architecture

From intention to a physical task.

In a planned laboratory workflow, a nurse prepares a tracked simulated sample and hands it to a courier. The courier carries it to the laboratory, where a scientist records receipt. Once the simulation produces a result, the responsible clinician is notified. These embodied workflows require permissions, object-state validation and confirmed arrival; the complete sequence is not implemented in the current prototype.

Prepare and hand over a tracked sample
Carry it to the laboratory and confirm arrival
Record receipt and notify the permitted clinical team

04 / Simulation architecture

A private intelligence runtime.

The current prototype uses local inference. Streaming speech recognition, visual perception and multilingual orchestration are development directions.

Local model runtime in the prototype
Speech synthesis connected to dialogue
Streaming speech input and visual context planned

A clear development boundary

Orthopedics is the initial prototype focus. Specialty scope, advanced physiology, institutional authoring and multiplayer describe the platform direction. No clinical validation or regulatory approval is claimed.

See what exists today

Architecture in context

A role comes with a boundary.

Explore how permitted knowledge and professional responsibilities shape a character's place in the medical world.

Concept studyIntelligent population · planned role and permission model

Persistent identity / Clinical

Nurse

Permitted context: Assigned patients, recorded observations and the nursing care context.
Role-specific task: Observe, record and hand over changes in the patient's condition.
Communication & escalation: Escalate concerns to the appropriate clinician.

A person inhabits a role, a care team and a continuing world. This explorer demonstrates the architecture; it is not a live AI conversation.

Architecture in context

An encounter becomes relevant memory.

Identity, role and permitted history shape the next conversation. Private deployment, memory controls and auditability need to be evaluated together.

Concept studyPatient continuity · illustrative planned workflow

Day 1 / Encounter memory

The first encounter.

Identity, relevant history and the initial encounter are attached to one patient record.

Memory has scope. A previous conversation does not grant access to every result or every team's record.

Questions for an evaluation

Scope the model before deployment.

Define the clinical task, available hardware, intended learners and required evidence before deciding which capabilities belong in an institutional evaluation.

01Can multiple clinicians train together?

Mixed human and AI teamwork is part of the design. A synchronized multiplayer experience is planned; the current prototype provides shared-backend foundations rather than a completed multi-user product.

02Can hospitals create custom scenarios?

Educator authoring is planned for patients, conditions, objectives, teams, complications and rubrics. It is intended to reduce reliance on Unreal source changes, but a production authoring interface is not yet available.

03Can AESORYX run privately or on-premises?

Local inference is used in the current prototype. Private cloud and on-premises delivery are deployment directions to assess with institutions; a production deployment package has not been established.

04Is AESORYX clinically validated or a medical device?

No clinical validation, regulatory approval or clinical decision-support status is claimed. The current work is a simulation prototype. Clinical judgment, patient care and regulated device use require separate governance and evidence.

For institutions building what comes next

Build the next generation
of medical simulation.

Start with your specialty, your team and a concrete learning or research objective.

Request a demo Contact AESORYX