Skip to main content

Platform Overview

This page is the map of Tietai: how data moves through the platform, where each area is in the sidebar, which page documents it, and how mature each part is.

How the pieces fit together​

Your systems send messages to Connect, Connect writes patient data to the FHIR server, Studio workflows and Agentic agents act on it, and Apps and Govern modules show and control the result

Read it left to right and match it to the sidebar:

  1. Your systems (hospital systems, devices) send messages.
  2. Tiet · Connect receives them. Network → Pipes handles message channels and Device Fleet handles wearables and devices.
  3. Patient data lives in the FHIR server, which you browse under Tiet · Govern → FHIR Server.
  4. Tiet · Studio workflows and Tiet · Agentic agents read and process that data. Agents can run as nodes inside workflows.
  5. Apps (clinical applications) and Tiet · Govern (knowledge, imaging, dataset publication) put the results in front of people and control access.

Every area is scoped to the active organization and gated by permissions. See Key Concepts.

The sidebar map​

The sidebar has six groups. An item appears only if your organization has the module and you hold one of its permissions, so your sidebar may be shorter.

GroupItemSub-items
Tiet · HubHomeOne entry per dashboard workspace of your organization
AnalyticsBoards
PlatformMonitor, Projects, Connectors, Egress policy, Environments
AppsTiet · Consult, Tiet · Scribe, Patient Screening, Voice Workflows, Medical AI / Clinical Reasoning, Tiet · Coding, Clinical Processes, Tiet · Documents
Tiet · ConnectNetworkMesh, Pipes, Messages & DLQ, Transformers
Device FleetDevices, Providers, Enrollment, Alerts, Automations, Settings
Tiet · StudioWorkflowsWorkflows, Monitor
Tiet · AgenticAgentsAgent Workspace, Agents, Monitor, Reviews
ComplianceSystems, Policies, Gate, Watch, CAPA, Audit
Tiet · GovernKnowledgeLibraries, Documents, Web, Query
FHIR ServerExplorer, Editor, Patients, Server
Records ViewerDICOM Studies, DICOM Viewer, ECG / Holter, Labs
EHDS StudioOverview, Publish Dataset, My Work, Datasets, Sources, Setup & Admin, Public Catalogue
SystemAdmin ConsoleOpens its own menu: Account, Organization, Platform, Security
Help & Support

The Agents page has its own tabs for Agents, Agent Graphs, Agent Catalog, Models, Tools and Skills.

What each area does​

AreaWhat it doesPage
HomeGreeting, one shortcut per enabled module, dashboards, monitor tiles, service health, recent workUsing the application
AnalyticsDashboards and BoardsAnalytics dashboard
PlatformProjects, Connectors, Egress policy, Environments, platform monitorTiet · Studio and Tiet · Agentic, Connector directory
AppsClinical applicationsTiet · Consult, Using the application
NetworkMessage pipes, mesh, dead-letter queue, transformersTiet · Connect, Run on Mesh
Device FleetWearables and device dataWearables and IoT
WorkflowsVisual workflow builder and run monitorTiet · Studio Builder, Environments and promotion
AgentsAgents, agent graphs, models, toolsAgents and governance, Create agents
ComplianceAI system inventory, policies, release gate, drift watch, auditAgents and governance
KnowledgeKnowledge libraries, documents, web sources, query testingNo dedicated page yet
FHIR ServerBrowse and edit FHIR resources, patients, server operationsFHIR and interoperability, Manage patients
EHDS StudioGoverned health-data catalogue recordsEHDS Studio
Admin ConsoleProfile, organization, users, roles, secrets, SSO, auditAdministration

A message becomes FHIR data​

The path below is the one you can build today with the workflow engine's own HL7 listener. An HL7 v2 message arrives on the workflow engine MLLP port (default 2575, set by MLLP_PORT). The workflow engine looks for saved workflows that contain an hl7-listener node on that port and runs them. A typical workflow chains hl7-to-fhir and fhir-crud.

A hospital system sends an HL7 message to the workflow engine MLLP server, which runs the saved workflow: hl7-to-fhir builds a transaction Bundle, fhir-crud posts it to the FHIR server, and an acknowledgment returns to the sender

Things that surprise people:

  • A message that no workflow listens for is rejected with AR by default (the MLLP_NO_ROUTE_ACK setting can change the code). A message that every listener filters out is acknowledged AA and ignored, unless the listener sets nack_on_filter (then AR).
  • hl7-to-fhir maps Patient and Encounter. For a test ADT^A01 message the visit number, next of kin, allergies and diagnoses were not mapped; the node lists unmapped segments in its result.
  • Every message creates new resources (the Bundle uses POST). Nothing is matched or updated.
  • Patient matching is not part of this path. The master patient index is a separate node (empi) that you add to a workflow yourself.
  • Connect pipes can also receive HL7 on a TCP/MLLP source. The capability matrix lists the TCP/MLLP listener as generally available.

Maturity at a glance​

See Capability status for the shared availability table, node maturity counts and verification dates. Check the documented limitations for the feature you plan to use.

Next steps​

Something unclear? Send documentation feedback by email.