Problem

Aluna addresses a bottleneck in biomedical AI: health organizations need trustworthy human-derived data and defensible evaluations to build and compare models, but assembling curated datasets and repeatable benchmarks is difficult, slow, and costly. The practical economic pain is the duplicated data-curation, specialist annotation, and evaluation effort that each organization would otherwise bear, along with the risk of investing in models whose performance cannot be compared reliably. The clearest killer use case is oncology AI, where teams need to evaluate systems across cancer types and treatment modalities rather than rely on ad hoc datasets or narrow claims.

Product / Service

Aluna positions itself as a data foundry that works with health organizations to build curated datasets and evaluations for health AI. Its offering centers on curated human data and oncology-focused benchmarking: OncoBench is described as a forthcoming suite with standardized datasets, evaluation metrics, and performance comparisons across cancer types and treatment modalities. This gives researchers and companies a more consistent basis for training, testing, and selecting biomedical AI models. Aluna’s public technical footprint also includes open oncology benchmark code, a whole-slide-image deidentifier, and a patient tokenizer, suggesting that its delivery model combines data products with supporting tooling.

Market

Aluna competes in life-science AI data infrastructure, specifically the market for biomedical datasets, oncology data, and AI evaluation and benchmarking. The evidence does not identify a single direct like-for-like rival, but nearby alternatives include Protege Data Lab’s healthcare evaluation datasets and benchmarks, Champions Oncology’s multi-omic data platform, and established pathology benchmarks such as CAMELYON16/17. These alternatives indicate that Aluna is entering a market where proprietary clinical data platforms, curated evaluation datasets, and open research benchmarks overlap.

Aluna is an early-stage company founded in 2024, part of Y Combinator’s Summer 2024 batch, listed as active, and based in San Francisco with a three-person team. Its GitHub organization shows active benchmark and data-processing repositories, but OncoBench is still marked “Coming Soon,” and a company profile labels Alunadata unfunded. The available evidence contains no disclosed customers, revenue, or financing, so the best-supported characterization is pre-commercial or pre-revenue, with product-building traction rather than demonstrated commercial traction.