Enterprise

Epic Unveils Agent Factory and Cosmos-Powered AI Predictions

The EHR giant positions itself as healthcare's AI infrastructure layer with no-code agent builder and clinical forecasting model.

Omega Editorial· August 19, 2026· 3 min read

Epic Unveils Agent Factory and Cosmos-Powered AI Predictions

Electronic health record leader Epic Systems is positioning itself as the foundational AI infrastructure for healthcare, announcing a suite of tools designed to embed artificial intelligence directly into clinical, operational, and administrative workflows. At its annual Users Group Meeting in Verona, Wisconsin this week, the company unveiled Agent Factory, a no-code platform for building AI agents, and Curiosity, a generative AI model trained on real-world patient data.

Why it matters

As healthcare organizations face pressure to adopt AI quickly while patients increasingly turn to consumer AI tools for health information, Epic's integrated approach could determine whether health systems maintain control over clinical decision-making or lose ground to third-party vendors and direct-to-consumer platforms. The company's bet on deep EHR integration contrasts with the fragmented marketplace of AI point solutions.

Agent Factory enables custom AI without coding

Agent Factory, previewed at the HIMSS26 conference earlier this year, allows healthcare organizations to build, run, and monitor AI agents that can reason and act across workflows without writing code. The platform offers 120 out-of-the-box AI features that organizations can customize or use as templates for new agents.

Early adopters are already deploying custom agents. ECU Health built an AI agent to summarize transfer center requests, improving handoffs and saving an estimated 20 hours weekly, according to Epic. The platform will become widely available in 2027.

Curiosity forecasts outcomes using 310 million patient records

Epic is leveraging Cosmos, its clinical research database containing data from over 310 million patients and 21 billion medical encounters, to power Curiosity. This generative AI model forecasts clinical outcomes such as hospital readmission and stroke risk using real-world patient data rather than relying solely on published studies.

Twenty healthcare organizations are currently validating Curiosity, and the predictive tool will be available to any organization contributing data to Cosmos. CEO Judy Faulkner said Curiosity can answer clinical questions at both population and individual patient levels, even when no published research exists.

Integrated AI ecosystem across the care continuum

Epic is positioning its AI tools—Art for clinicians, Emmie for patients, and Penny for revenue cycle—as components of a unified enterprise system called Ergo. The company describes Ergo as "healthcare intelligence" that combines human judgment with AI's pattern recognition across massive EHR datasets.

Ergo Visit, now live at Ochsner Health and five physicians in Epic's Garden Plot collaborative, demonstrates this integration. Before appointments, Art summarizes patient records while Emmie gathers information from patients through MyChart. During visits, Chart with Art listens to conversations, drafts notes, and prepares documentation. More than 300 healthcare organizations currently use Art's summary capabilities.

Urgency around patient adoption of consumer AI

Carl Dvorak, a top Epic executive, emphasized that patients are rapidly adopting AI tools independently, creating pressure on health systems to move faster. He warned that healthcare organizations risk losing trust as patients' primary source of medical information if they cannot keep pace with AI innovation.

"We've got to find a pathway to our AI utopia before patients become frustrated with the health system," Dvorak told attendees, urging healthcare leaders to engage with regulatory bodies and lead internally rather than delay adoption.

These details were first reported by Fierce Healthcare, which covered Epic's Users Group Meeting announcements.

#epic systems#healthcare ai#ehr#clinical decision support#predictive analytics#agent factory

This is an original analysis by the Omega editorial team. Source reporting: Automation Watch.

Want systems like this working for your business?

Book a Call

More in Enterprise

Enterprise· 3 min read

AI Medical Tools Fail Early Diagnosis Steps Despite Strong Outcomes

New research shows frontier AI models excel at naming diseases but struggle with the differential diagnosis that prevents dangerous errors.

Via AI Watch · Aug 19, 2026
Enterprise· 3 min read

AI Receptionist Struggles With Yorkshire Accents at UK Clinics

Healthwatch Rotherham reports patients face barriers with EMMA system that handles appointment bookings at GP surgeries.

Via AI Watch · Aug 19, 2026
Enterprise· 3 min read

Hims & Hers CEO: Open-Weight AI Models Cut Costs 70-80%

Andrew Dudum says companies with proprietary datasets should train their own models rather than rely on commercial AI services.

Via AI Watch · Aug 19, 2026