Trellis AI (YC W24) is hiring engineers to build AI agents for healthcare access

Robots vs Red Tape: Will AI speed meds—or gatekeep care

TLDR: Trellis AI is hiring to build bots that automate insurance approvals and billing so patients get treated faster. Commenters are split: clinicians cheer an end to fax hell, while skeptics worry “AI coworkers” means denials, layoffs, and privacy risks—debating whether this kills red tape or automates gatekeeping.

YC newcomer Trellis AI says it’s hiring builders for “AI coworkers” to battle the worst part of U.S. healthcare: prior authorizations—the insurer permissions that hold up treatment. The pitch: bots that read messy chart notes, file appeals, and shove paperwork through so fast some clients saw treatment times drop by 90%. The comments? Pure chaos—and comedy.

Clinicians and office admins piled in with exhausted cheers: kill the fax, save my lunch break. One nurse said if a bot handles prior auths at 3 a.m., “that’s two more chemo chairs tomorrow.” But a loud skeptical camp heard another story. “Stripe of healthcare?” they snorted, dubbing it the Ticketmaster of approvals and side‑eyeing the mysterious “XX% market share.”

The phrase “agents that make critical decisions” lit a fuse. Privacy hawks asked who audits the bot when it says no. Engineers roasted the buzzwords—“agentic frameworks” got labeled glorified forms plus a chatbot—while others argued robust testing could actually make this safe at scale. Labor worries spiked too: “24/7 AI coworkers” read as code for layoffs.

Memes flew: Clippy in scrubs, “Captcha for chemo,” and a prior auth speedrun% any%. Yet beneath the snark, a rare unity: everyone hates the paperwork boss fight. If Trellis melts the red tape, it’s a hero; if it automates the gatekeeping, expect pitchforks

Key Points

  • Trellis AI builds computer-use AI agents that automate healthcare document intake, prior authorizations, and appeals, processing billions of dollars in therapies annually across all 50 U.S. states.
  • The company is a Stanford AI Lab spinout backed by Y Combinator, General Catalyst, Telesoft Partners, and executives at Google and Salesforce.
  • The role involves designing agentic frameworks for healthcare decision-making, deploying 24/7 long-running agents, and delivering production-grade systems with a comprehensive evaluation suite.
  • Required skills include Python or Go, ML/NLP libraries (PyTorch, TensorFlow, Transformers), databases (especially Postgres), and experience with cloud (AWS, Azure, GCP) and containerization (Docker, Kubernetes).
  • Trellis reports outcomes such as >90% reduction in time to treatment, improved prior authorization and reimbursement rates, and conversion of unstructured documents into structured EHR data; it also reports 10x revenue growth recently.

Hottest takes

"Call it what you want—if it kills fax machines, I’m in" — scrubsAndSnacks
"‘Stripe of healthcare’ or Ticketmaster of prior auth?" — copayCowboy
"Who audits the bot when it denies grandma’s meds?" — HIPAAHawk
Made with <3 by @siedrix and @shesho from CDMX. Powered by Forge&Hive.