Amid tech’s AI push, some health systems forge their own path

The new year began with a bevy of long-awaited announcements from technology companies making inroads in artificial intelligence for healthcare. However, that hasn’t stopped large health systems from pursuing their own AI efforts. 

In the first weeks of 2026, OpenAI launched ChatGPT Health, Anthropic introduced Claude for Healthcare, and Amazon One Medical launched its Health AI Assistant. The OpenAI and Amazon releases largely target patients, promising personalized health guidance based on aggregated medical records coupled with wellness data. Meanwhile, Anthropic is emphasizing its potential to automate administrative workflows for health systems. 

These releases put the industry on notice that AI innovation is here. It’s not just tech giants making waves, either: Research from Bessmer Venture Partners found AI companies got 55% of healthcare tech’s $14-plus billion in funding in 2025. 

Leading hospitals and health systems haven’t taken the news sitting down. Many have been developing their own AI tools, some leveraging internal expertise and others partnering with startups.  

• Cleveland Clinic has built AI tools alongside numerous partners, as Healthcare Brew detailed. Products include an ambient scribe, revenue cycle management (RCM) coding, and sepsis prediction. The health system takes this approach, according to Chief Digital Officer Rohit Chandra, because it can address specific problems.  

• Duke Health and Trase Systems are working on AI agents, starting in the Duke Heart Center. The goals of the partnership represent a common theme: Surface insights from longitudinal data sets – patients’ medical records, plus lifestyle and biology data – to inform personalized care and also improve efficiency. 

• Mass General Brigham spun out AIwithCare, which built a product to screen patients for eligibility in clinical trials. RECTIFIER, uses retrieval-augmented generation to access unstructured electronic health record (EHR) data and identify patients who meet a trial’s eligibility criteria. Two research studies found RECTIFIER identifies patients faster and more accurately than manual methods. The tool now supports roughly two dozen use cases in research and clinical operations. 

• Penn Medicine created Chart Hero, a “sidebar” in the EHR for clinicians to “gather, arrange, synthesize, and assist in the interpretation of all the pertinent information they’d need for a patient they’re going to see.” The health system takes pride in developing a product based on a diverse data set and with clinical context often missing from commercial AI models. 

• Stanford Medicine developed ChatEHR to similarly help clinicians quickly gather information from a patient’s record and ask relevant follow-up questions – valuable in scenarios when reviewing an entire chart takes far too much time. Researchers in the health system spent two years developing the tool and tested it with a small cohort before a broader rollout. 

Health systems have multiple motivations for rolling their own AI products, according to McKinsey. Their data sets offer a foundation that’s solid and trustworthy, and they can co-develop alongside clinical champions instead of presenting them with fully baked but potentially suboptimal tools. Successful tools can have financial value as commercial products, too. Plus, for all the noise big tech made with its healthcare AI announcements, it remains to be seen whether patient-facing AI will fare better than personal health records.  

For organizations hoping to use AI in their own ways, the American Hospital Association offers advice for drafting an AI action plan. The report cites nine building blocks within the common pillars of people, process, and technology, emphasizing the importance of leadership, change management, centralized capabilities, data governance, and infrastructure suitable for AI workloads. 

AHA noted there’s no single path forward. Some organizations address one use case at a time, while others pursue several at once. Nor is there a universal starting point, as there are opportunities within administrative, clinical, financial, and operational workflows. The most important takeaway? “The time to start is now.” 

Brian Eastwood is a Boston-based writer with more than 10 years of experience covering healthcare IT and healthcare delivery. He also writes about enterprise IT, consumer technology, and corporate leadership.