OpenAI President Greg Brockman: Building the Best Healthcare Platform
OpenAI President and co-founder Greg Brockman stated that the company is working to establish OpenAI as the world's best healthcare platform, ushering healthcare into the era of artificial intelligence.
He pointed out that patients must carry their medical records and repeatedly explain their conditions when consulting different specialists, ultimately leaving the decision-making to the doctors alone; if all parties are on the same platform, medical records and examinations can truly flow across institutions. He identified patient recruitment for clinical trials as a significant bottleneck in new drug development, emphasizing that complete data and users willing to share their information are essential for quickly finding eligible participants who may benefit.
OpenAI's healthcare business operates along three lines: consumer-facing health Q&A, which sees hundreds of millions of queries weekly; specialized dialogue and literature citation tools for clinical doctors; and enterprise delivery for hospitals, which has already integrated with Epic electronic health records. OpenAI for Healthcare and ChatGPT for Healthcare have been launched at institutions such as AdventHealth, Cedars-Sinai, HCA, Memorial Sloan Kettering, and UCSF, providing HIPAA-compliant business agreements.
The company has also partnered with Massive Bio to parameterize and conduct real-time pre-screening of trial inclusion criteria; additionally, it has launched a search plugin for public datasets like ClinicalTrials.gov, PubMed, and DailyMed. A collaborative study with Penda Health in Kenya covered approximately 40,000 visits, reporting a relative decrease in diagnostic and treatment errors. The physician review network spans dozens of countries and multiple specialties.
He articulated the platform's value as: the same set of records can reduce the referral burden on patients and shorten trial matching times. The premise is that users must entrust their health data to a single technology platform.
In terms of market mechanisms, this is a competition for data entry: patients and doctors provide dialogue and medical record summaries to the model, while hospitals grant electronic health record query rights to the interface. Beneficiaries include model companies that have integrated with mainstream electronic health records and can serve both consumer and hospital sides, as well as trial service providers that charge for matching participants. The pressured parties are decentralized specialty information systems, manually screened contract research organizations, and hospitals reluctant to entrust longitudinal records to a single cloud. Funding is shifting from fee-for-service software licenses to subscriptions based on retained health data and workflows.
Brockman emphasized that only when people are willing to share their information can the platform serve both individuals and drug development. Medical licensing, liability attribution, and interstate data regulations will still determine how far this narrative can go.
Source: Public Information
ABAB AI Insight
OpenAI first builds weekly active scale through consumer health Q&A, then sells the same model to doctors and hospitals, a path similar to its programming assistant: first free entry, then embedding workflows into professional systems. Brockman identifies "carrying medical records and repeated statements" as product defects, with the solution being that all specialties stand on the same platform. Clinical trial recruitment is the second monetization: once medical records are searchable, participant matching shifts from manual record reading to model pre-screening. The Epic interface is a key gateway; without the context of electronic health records, the hospital side is just another chat window.
Capital is shifting from training computing power to compliance and integration. HIPAA business agreements, physician review benchmarks like HealthBench, and public data plugins are all aimed at enabling procurement departments to sign off. Collaborations like Massive Bio turn trial protocols into machine-readable parameters, indicating that pharmaceutical budgets may monetize faster than patient subscriptions. The motivation is that health inquiries are already happening in ChatGPT, and the strategy is to upgrade these dialogues into auditable clinical record layers.
In contrast to Google’s long-term binding with electronic health record vendors, Epic’s own application market, and Tempus and Flatiron selling oncology data to pharmaceutical companies, the industry position is shifting from auxiliary Q&A to control layers: whoever maintains the cross-specialty timeline can charge both hospitals and sponsors.
This represents a transfer of pricing power: the value of medical information shifts from single visit records to continuously accessible platform files. The mechanism is that decentralized systems force patients to act as transporters, while a unified platform collects transport costs as subscriptions, and trial recruitment allows the same record to be sold again.
ABAB News · Cognitive Laws
Whoever reduces the need for patients to explain once collects the medical record entry
The bottleneck in trial recruitment is not the drug, but the inability to find qualified individuals
Once the platform holds health data, specialty systems become plugins