Tennr uses vision-language models and document AI to automate healthcare patient referrals, from intake through prior authorization, processing over 10 million documents a month for US providers.
Tennr builds AI systems for one of American healthcare's most stubborn operational problems: the patient referral. Roughly 90% of provider-to-provider communication still runs on fax, and referrals arrive as faxes, handwritten notes and PDFs of 20 pages or more in thousands of inconsistent formats. Tennr's answer is RaeLM™, a vision-language model trained on more than 3 million medical documents that reads referral paperwork, extracts what matters and drives the workflow from intake through prior authorization. The engineering problems are concrete: document understanding at scale, information extraction from messy real-world inputs, and tight integration with EHR systems.
The referral automation platform built on top classifies documents, populates EHRs without human intervention, verifies benefits and handles the correspondence that typically stalls care. It processes more than 10 million documents a month for healthcare providers across the United States. The stated outcomes are operational rather than aspirational: capturing more referrals, reducing first-pass denials and moving patients from fax to first visit faster.
Technical work spans vision-language models, document AI, natural language processing, machine learning, workflow automation and healthcare IT. Tennr treats the messiness of the domain - faxes, handwriting, non-standard formats - as a design constraint rather than an exception, and its stated mission is to close the referral "black hole" so patients are not lost between providers. The company communicates with unusual candor about the state of the industry it serves.
