Qventus

Prediction and automated action are the product. Qventus describes a system of action on top of the system of record: the platform pulls data streams from the record, predicts the barriers that will hold up a patient or a case, and takes the next step for the care team, populating expected discharge dates and dispositions, prioritizing ancillary orders, following up pre admission testing, and releasing and filling operating room time.

Remove the models and there is nothing left to sell. There is no bed board, scheduling system or reporting layer underneath that a customer would still be buying; the forecast is the input and the action is the output. Models are trained on each customer's own data, adapted to its patient population and tuned to its processes, which makes them more central still.

The assistants act on their own inside operational work. They read, gather, understand and write, anticipate the next step and act for the care team across discharge planning coordination, ancillary order prioritization, document and fax handling and pre admission testing follow up. The boundary is drawn by kind of work: clinical judgment stays with the care team, and outputs such as an expected discharge date in the chart are forecasts a clinician can override.

What an assistant does when it is unsure, what routes a case back to a person, who hears when an automated action turns out wrong, and how a customer sees or sets the thresholds of a model tuned to its own hospital are not described. For discharge coordination that matters even when the work is administrative, since the sequence of orders and the planned date shape when a patient actually leaves.

The platform combines real time data with machine learning, generative AI and behavioral science, and its AI operational assistants read, gather and write inside the record. Models are trained on each customer's own data, adapted to its patient population and tuned to its processes rather than shipped as one generic model. OpenAI appears on the company's subprocessor list, the one AI model company named there.

No model or model family is identified for any capability, and no accuracy figure or update practice is published.

Nine subprocessors are listed in the trust center: Amazon Web Services as cloud provider, Databricks and Snowflake for data storage and processing, OpenAI and Twilio for engineering, Google Workspace for identity, and GitHub, Datadog and Notion. OpenAI is the one AI model company on the list, with its role described as engineering, and the website privacy policy reserves an indefinite right for Qventus and its business partners to use and disclose de identified data.

Which assistants call OpenAI's models, and on what data, is not stated.

Named deployments carry dated figures. HonorHealth, a six hospital system in Arizona running Epic, started in 2021, and its story reports 181 beds' worth of capacity created by reducing excess days, $69 million in savings, 224 incremental perioperative cases a month and a 9.6 times return, with a 7 percent rise in completion of high priority orders described as correlating with 0.4 fewer excess days. OhioHealth, Northwestern Medicine, Jackson Health, Boston Medical Center and Allina Health appear in further stories, with operating room hours unlocked and multi year results, and aggregate client figures for 2025 are published alongside them.

All of these figures are Qventus's own, and the link between order completion and excess days is stated as a correlation; the KLAS recognition the company cites measures customer experience rather than prediction accuracy. Readmission rates over the same periods, the natural counterweight to excess days saved, are not reported.

Protected health information is the platform's working material. The trust center lists personal health information among the data collected, customer data is deleted when a customer leaves, and a HIPAA examination report and the security policies behind it are released on request.

How long data is kept during a contract, whether it trains anything beyond the customer's own models, and how data is de identified are not set out. The website privacy policy reserves the right for Qventus and its business partners to keep using and disclosing de identified data indefinitely, without defining the term. Whether that clause reaches platform data or only the website is unclear, so whether anything learned at one health system can leave it is not answered either way.

Every Qventus customer is a United States health system and the platform works on patient level data inside the record, so a business associate agreement sits behind every deployment. The trust center lists HIPAA as a compliance framework and releases, on request, a 2025 HIPAA Type 1 examination report, a HIPAA compliance policy and an incident response plan with a HIPAA breach notification addendum.

No agreement template is published, and the website privacy policy and terms of use cover the site rather than the service.

A SOC 2 Type 2 report for 2025 heads the trust center's documents, alongside a HIPAA Type 1 report, external network and web and API penetration test reports from October 2025, an architecture diagram and a disaster recovery exercise report. Listed controls include encryption, restricted production access, employee background checks and cybersecurity insurance. The reports are released on request rather than open to a visitor.

The platform is embedded in the record system across a large hospital estate and takes automated action on administrative workflows, and which systems and environments the SOC 2 report covers is not shown on the open page.

Qventus automates operational and administrative coordination, capacity, flow, discharge planning, perioperative preparation and operating room use, rather than informing diagnosis or treatment, and it does not present any product as a medical device.

The discharge products produce operational forecasts a clinician can override: a predicted discharge date shapes when a bed is freed, when ancillary orders are prioritized and when planning starts. The care gap and coding suite and the malnutrition module sit closer to clinical content, since they surface conditions for documentation and drive consults. How their outputs are labeled to clinicians is not described.

These models allocate scarce hospital capacity. A discharge readiness prediction, a priority order on an ancillary service, a decision about which case fills a released operating room block: each distributes access to care, and a systematic tilt in any of them would never surface as a clinical decision anyone reviews. A discharge date set slightly earlier for one group and later for another produces a real difference in length of stay that no chart audit would attribute to a model.

Training on each customer's own data means each model learns that hospital's historical patterns, including any inequities in them, with no external benchmark unless Qventus builds one. No governance framework, evaluation method, monitoring by payer, language, race or admission source, drift response or statement of what the models optimize for is published.

Qventus's assistants act inside hospital workflows: they set expected discharge dates and dispositions, prioritize ancillary orders, follow up on pre admission testing and release and fill operating room time, and clinicians can reverse an automated action. A wrong output shows up as a patient moved before the plan is ready, a bed held too long or a case unready on the day, rather than as an obvious fault.

No accuracy figure, warranty, service level or remediation commitment is published, and the website terms of use cover only the site.

The record system is where this product lives. The platform pulls several data streams from it and acts inside it: the inpatient product populates expected discharge dates and dispositions in the chart by the first morning after admission, which is structured writing back, and the perioperative and surgical products act the same way on case scheduling and operating room time. HonorHealth's six Epic hospitals have run it since 2021, which puts it in production on the largest record system in the market.

No integration documentation, marketplace listing, interface standard or other record system is named, so support for a second record system in production, and what the write back consists of, is not shown.

The platform runs on Amazon Web Services, which Qventus's trust center names as its cloud provider, with Databricks and Snowflake for data storage and processing and a disaster recovery exercise report among the documents released on request. Models are trained per customer, which protects one health system's data only if derived artifacts stay isolated per customer.

The AWS regions, tenancy model and any residency commitment for health system data are not stated. The website privacy policy's promise that site data stays in the United States covers the website rather than the platform.

Qventus sells by solution line: inpatient capacity and discharge, perioperative care coordination, surgical growth and operating room utilization, a care gap and coding suite and a malnutrition module, with an AI Solution Factory for custom workflows. Contracts are reported to be custom and scoped to each health system, and the route in is a demo request. Deployments include record system integration and change management, so a services component sits inside each contract.

No price, tier, unit of charge, fee or term is published. Qventus does publish specific returns at named health systems, including multiples of annual return and dollars saved, so the value side of a business case can be modeled from public material while the cost side waits for a sales conversation.

Coverage is deliberately scoped to acute care operations and named by care area. The solution lines are inpatient capacity and discharge, perioperative care coordination built around pre admission testing, surgical growth and operating room utilization, a care gap and coding suite aimed at earlier intervention and documentation of complex conditions, and a malnutrition care module, used across more than 150 hospital facilities. Buyer pages address surgical services, inpatient care and system or facility leadership.

The perioperative line also sells administrative work in its own right. Intelligent Document Management processes faxed, scanned and emailed records automatically, and the company reports manual fax processing time down about 60 percent, to roughly three minutes a fax; PAT Assist handles records requests, chart retrieval and note taking; and the AI Patient Concierge contacts patients by voice, text and email to gather intake, coordinate appointments and track follow up.

Named customers span academic medical centers, community and regional systems and multi hospital groups, from HonorHealth's six Epic facilities to Northwestern Medicine, OhioHealth, Jackson Health, Boston Medical Center and Allina Health. The care gap, coding and malnutrition modules carry one named site between them. Ambulatory, post acute and payer settings are outside scope by design.