Trellis AI, a Stanford AI Lab spinout, builds healthcare-native AI agents that automate referrals, benefits verification, prior authorisation and reimbursement workflows for specialty pharmacies and infusion providers.
Trellis AI is a spinout from the Stanford AI Lab building healthcare-native AI agents that automate the administrative machinery surrounding specialty medicines and therapies. The company's focus is on the document-heavy processes that determine whether patients gain access to high-cost therapies - referrals, chart notes, benefits verification, prior authorisation - starting with specialty pharmacies and infusion providers. For engineers, the appeal is the nature of the problem: fragmented, unstructured, and consequential, with the surrounding workflows covering healthcare operations involving patients across all fifty US states.
The technical work spans clinical document extraction, natural language processing applied to referrals and chart notes, process mining, insurance benefits verification automation, and workflow automation. The Trellis AI Platform extracts clinical information from referrals and chart notes, verifies medical and pharmacy benefits, investigates reimbursement, generates letters of medical necessity, submits prior authorisations, manages appeals, and monitors payer and patient status. Three capabilities structure the system: Copy, which maps existing business processes within healthcare organisations; Capture, which turns proprietary data into executable workflows; and Execute, which uses AI agents to perform and verify individual workflow steps.
The stated outcomes are operational rather than aspirational: increased referral conversion, reduced bad debt, improved reimbursement, and support for higher patient volume without proportional staffing growth. The company operates across healthcare, specialty pharmacy, infusion services, and revenue-cycle administration in the United States.
