January 6, 2026
Pricing, pipelines, and pharma drama
Launch HN: Tamarind Bio (YC W24) – AI Inference Provider for Drug Discovery
HN loves the cure tech, drags 'book a meeting' pricing and 'another scheduler'
TLDR: Tamarind Bio launched a platform that lets pharma run powerful AI tools for designing drugs without heavy setup. Commenters cheered the mission but clashed over “book a meeting” pricing, questioned building a new scheduler, and pressed on open‑source vs proprietary models—making clear that usability and transparency will decide its impact.
Tamarind Bio rolled up to Hacker News with a bold pitch: a one‑stop shop to run heavy‑duty AI for drug design, from protein tools like AlphaFold to custom pipelines—all wrapped in a friendly web app so non‑coders can play scientist. They claim much of Big Pharma is already on Tamarind, and there’s a slick demo to prove it. Cue the community chorus. The loudest clapback? Pricing. Commenters groaned at the dreaded “book a meeting” wall, with one basically begging, just show a number already. Others hit the brakes on the tech: a dev from an established scheduler project dropped in to ask why Tamarind built yet another job scheduler instead of using existing tools. Spicy!
Curiosity ran high too. Folks wanted to know how much Big Pharma really trusts open‑source models versus their closely guarded in‑house ones, and whether Tamarind can handle both. A product‑focused commenter asked what “upload a protein file and generate molecules” actually does under the hood—and how long it takes—because that’s a broad promise. And the “can you Benchling your way into pharma?” question came up, probing how Tamarind won over big, slow‑moving companies.
The vibe: optimistic about faster drug discovery, skeptical about vendor opacity, and amused that even miracle‑drug dreams still trip on the ancient boss battle of enterprise software—calendars, contracts, and “we built our own scheduler.”
Key Points
- •Tamarind Bio provides an AI inference platform for drug discovery, serving open-source models like AlphaFold via APIs and a web app for non-technical scientists.
- •The platform implements standardized data schemas and a custom scheduler/queue optimized for long GPU-bound inference jobs and horizontal scaling.
- •Customers can chain multiple models into reproducible pipelines, onboard proprietary models, fine-tune, build UIs for Docker containers, and connect to wet lab data sources.
- •Tamarind reports usage by much of the top 20 pharma, dozens of biotechs, and tens of thousands of scientists, with some deprecating internal solutions to adopt Tamarind.
- •The company originated from bottlenecks observed in a Stanford lab and similar challenges within pharma, aiming to centralize and simplify computational biology workflows.