PillPack co-founder Elliot Cohen launches General Medicine, a chat-first healthcare marketplace with full health-history awareness
Dec 16, 2025 with Elliot Cohen
Key Points
- General Medicine, founded by PillPack co-founder Elliot Cohen, launches a chat interface that automatically imports 20+ years of patient records, enabling AI to connect current symptoms to historical medical events before clinical visits.
- The platform operates as a licensed provider group rather than a consumer app, allowing it to aggregate records more comprehensively than standalone import tools and match patients to specialists based on procedure-level experience.
- Cohen argues US healthcare's core problem is that patients are not the actual customer, and positions General Medicine to rebuild around patient experience as the primary metric, with AI as the mechanism to shift power away from insurers and employers.
Summary
Elliot Cohen, co-founder of PillPack, which sold to Amazon in 2018, is back in healthcare with General Medicine, a provider group that launched in May 2025 and has since expanded to nationwide availability. Cohen claims it carries broader insurance coverage than any other provider group in the country, and can route patients to local or specialist care when it falls outside its own network.
The company's core product announcement is a chat interface that is fully health-history aware, directly addressing the gap Cohen identifies in consumer AI usage. He estimates 60 to 70% of adults are already using tools like ChatGPT or Gemini to explore health questions, but those conversations are disconnected from actual care pathways. General Medicine's chat is integrated with patient records and feeds directly into clinical visits.
Record aggregation is the structural differentiator. Because General Medicine operates as a licensed provider group rather than a consumer app, it can pull records more comprehensively than one-off import tools. Cohen reports that when he loaded his own records, the system surfaced more than 20 years of history automatically, going back to 2005, which enabled an AI-assisted connection between a current two-year foot pain issue and an old ankle injury.
The clinical workflow Cohen describes positions the chat layer as pre-visit preparation rather than a diagnostic substitute. Conversation summaries are shared with clinicians ahead of appointments, allowing a 15 to 20 minute visit to skip the intake ramp and move directly to treatment decisions. In his own case, that process led to an MRI that identified ankle cysts, with the AI accurately interpreting the radiology report, verified by a radiologist friend.
The marketplace model runs on two tracks. Clinicians can join the platform directly and bill for care as they would within any health system. Separately, General Medicine maintains a database of physician specialties and procedure-level experience, allowing it to match patients to the precise specialist for a given need, not just a general category. Cohen frames this as aligning patient and physician incentives: patients reach the right provider faster, and physicians concentrate on the case types where they perform best.
On the broader question of measuring AI's impact on healthcare, Cohen argues current metrics are built around the financial relationship between providers and systems rather than patient experience. His preferred KPI is Net Promoter Score for healthcare interactions, a metric he acknowledges is currently near zero. His thesis is that most systemic failures in US healthcare trace to a single structural problem: the consumer is not the actual customer. The employer, insurer, or government holds that role. General Medicine's stated ambition is to rebuild around the patient as customer, with AI as the mechanism for shifting that power.