One intelligent operating layer for the whole hospital.

One comprehensive platform for the whole hospital — ICU to OR to ER to the back office — on one record and one audit log. It's built around the devices you already own and the way clinicians actually work, the AI is explainable and owned by your hospital, it runs in the cloud or fully on-premise, and it goes live fast. zMed has touched the lives of 300,000+ critical-care patients.

ICU and critical care, the operating room and anaesthesia, emergency/ER, ambulance and pre-hospital, tele-ICU and command centre, the hospital information system and ERP, clinical decision support, and the Claw AI layer — all on one platform, one record, one audit log.

zMed is comprehensive, and you choose how far to take it — three levels you can start at and grow through. Clarity automates charting and gives the whole unit live visibility. Insight adds continuous clinical scores, smart alerts, and analytics. Foresight adds predictive AI — deterioration and sepsis foresight, and demand and occupancy forecasting. Start at any level and move up when you're ready.

zMed is configured to your unit, not the other way round. Charting and protocols are built on zForms — configurable forms your hospital owns and changes the day a protocol changes, with no engineering involvement. Multiple zForms combine into a complete workflow for each unit (ICU, OR, ER, ward, and more), so the screen matches how your team already works. Templates, pathways, and rules are hospital-owned and adjustable per unit, role, and shift.

Three ways: zMed Cloud for units of 5–100 beds (SaaS, pay per patient-day, live in weeks); Enterprise for 100+ beds (on-premise or managed cloud, integrated with your existing systems); and a Tele-ICU & Command Center model for physician groups and multi-hospital networks.

Yes. You can start a self-serve trial at launch.zmed.tech, and we run on-site pilots for enterprise hospitals. Cloud units typically go live in weeks, not months.

zMed gives clinicians their time back — teams chart by exception instead of re-keying, and smart alarm suppression cuts alarm load by 60–85%. It surfaces deterioration and sepsis 4–6 hours before they're clinically obvious, so teams intervene earlier. And it replaces two systems with one — clinical and administrative on a single record, with charge capture driven straight from the chart.

Yes. zMed is device-agnostic and captures data directly from the monitors, ventilators, and anaesthesia machines you already own. It exchanges data with existing hospital systems over standard interfaces including HL7 and FHIR R4, so it fits the estate you have rather than replacing it.

Through zMed's medical device data integration (MDDI) layer, which connects directly to patient monitors, ventilators, infusion pumps, dialysis machines, and anaesthesia machines and streams their data — including waveforms — into the chart in real time. It's manufacturer-agnostic, so it works with the mix of devices you already own. For older devices that have only a serial port and no network, zMed provides serial-to-WiFi IoT adapters that bring them onto the network, so even legacy equipment streams into the chart — nothing re-keyed, every value traceable to source.

Yes. zMed exchanges orders and results with your existing laboratory and radiology systems over standard interfaces (HL7 and FHIR; DICOM for imaging). A lab order placed in the chart reaches your LIS, and results return to the ordering clinician in the chart with abnormal values flagged. Radiology reports live in the chart with the image a tap away, opened in zMed's built-in DICOM viewer; for hospitals that need it, zMed includes an integrated imaging archive (PACS) powered by Orthanc, so studies are stored and viewed right alongside the record. Adopt zMed's HIS & ERP and lab worklists, results, and pharmacy are native too — but zMed runs just as well alongside the lab and PACS you already have.

Claw is zMed's AI layer, built into the chart rather than bolted on as a chatbot. It drafts notes and summaries from voice, surfaces deterioration and sepsis foresight inline, and offers decision support at the point of care. Every suggestion is explainable, advisory only, and human-in-the-loop — the clinician is always in control.

The Claw layer spans prediction, automation, and assistance — built into the chart, advisory only, and human-in-the-loop (HITL): deterioration prediction (a continuous risk score and tier for every inpatient); sepsis foresight 4–6 hours pre-clinical; mortality and readmission risk; clinical-instability and smart-alarm pattern detection (alarm load down 60–85%); demand and occupancy forecasting; voice charting; self-drafting summaries (operative notes, shift handovers, discharge); and inline decision support. Every model is fitted on your hospital's own data, shows its contributing features, is one tap from override, and is audit-logged — the clinician's judgement is always the default.

A comprehensive, hospital-owned CDS library — around 500 rules at go-live (about 50 literature-cited), surfaced inline at the point of decision and growing over time. Areas include: sepsis bundles and severe-sepsis escalation; drug–drug and drug–allergy interaction checks; renal dose adjustment; VTE prophylaxis; early-warning and deterioration escalation; care bundles and order-set pathways for sepsis, stroke, trauma, chest pain/STEMI, and medico-legal cases across ICU, OR, ED, and the ward; continuously recalculated clinical scores (SOFA, APACHE II, GCS, NEWS2, MEWS, qSOFA, RASS, CAM-ICU, and more); documentation prompts and medication sign-off checks; and predictive models for deterioration, sepsis trajectory, mortality, readmission, and instability. Every suggestion is explainable, advisory only, and human-in-the-loop, and your team owns the rules through a visual editor — every accept, override, and defer is logged.

Built-in analytics and dashboards across clinical, operational, and financial views — census, throughput, length of stay, bundle compliance, turnaround times, and more — drawn live from the same record. Reports are template-driven and audit-linked, produced on demand or scheduled, and exportable to PDF for the board, payers, or regulators — every figure traceable back to the chart entry it came from.

Yes. zMed runs in the cloud or fully on-premise, and the AI runs in a secure environment inside your own infrastructure — so patient data never has to leave the hospital, including in-country deployments for data-residency requirements.

Privacy is the cornerstone of zMed's AI. Models run in a secure environment and are fitted on your hospital's own data, and your patient data is never used to train shared or third-party models. When a request is sent to a cloud AI model (such as Claude or OpenAI), protected health information is first redacted and de-identified by automated PII-detection tooling, so patient identifiers never leave your environment. Every prediction is advisory only and audit-logged, access is role-based, and deployments are HIPAA-aligned with data-residency options — including fully on-premise AI where nothing leaves the hospital at all.

The Claw AI layer and clinical decision support engine are explainable, advisory only, and human-in-the-loop (HITL) — the clinician is always in control, every suggestion shows its reasoning and can be overridden, and everything is audit-logged. The platform is designed to be FDA- and CDSCO-ready.

zMed is built for regulated healthcare environments, with HIPAA-aligned controls, role-based access, full audit logging, and data-residency options for on-premise or in-country deployments. Every record and every AI suggestion is audit-logged.