The scientific record, one API call away.
The science your models need does not live on the open web. It lives in behind paywalls in peer-reviewed journals, structured databases, and patent filings, in formats built for human subscribers. Valyu is the retrieval layer that hands it to your agents: literature, chemistry, genomics, clinical trials, and patents, full text, every result cited.
Valyu helps biotech and pharma teams move faster.
Peer review is the authority layer of science. When an AI system grounds an answer in a peer-reviewed article, it stands on that scrutiny. When it grounds an answer in whatever it could scrape, it is guessing. Preprints and abstracts get an agent to the edge of a question. The methods, the figures, the assay values a scientist would actually cite sit one paywall away. We close that gap - one call, full text, commercially clean, every claim traceable to its source.
Discover howBuilt for the way models actually read.
Response rate improved to 71% at 12 months.1
Name the sources, ask in plain English, get cited full text back.
One API across seven layers of biomedical knowledge — every result returns cited full text your agent can use.
| Source | Cited | |
|---|---|---|
| RETURNED | ||
| Neoadjuvant immunotherapy in resectable NSCLC | Wiley HLS | 312 |
| CAR-T persistence in relapsed B-cell lymphoma | Wiley HLS | 204 |
| Tau propagation in early Alzheimer’s disease | Wiley HLS | 158 |
| GLP-1 receptor agonists and renal outcomes | Wiley HLS | 96 |
| Microbiome shifts after fecal transplantation | Wiley HLS | 74 |
| Senolytics in age-related fibrosis | Wiley HLS | 61 |
| mRNA vaccine durability against variants | Wiley HLS | 48 |
| Ketamine for treatment-resistant depression | Wiley HLS | 33 |
Full text, one call away.
Run the whole workflow with DeepResearch — plan, retrieve across dozens of sources, verify, cite, and hand back a finished report. REST, MCP (mcp.valyu.ai), and plugins for Claude Code, Codex, Gemini CLI, and the Vercel AI SDK. Works everywhere you build. Three endpoints, one knowledge layer — Search, Answer, and DeepResearch.
See the APIs →
from valyu import Valyu
valyu = Valyu(api_key="val_***")
response = valyu.search(
"CAR-T cell therapy resistance mechanisms in relapsed lymphoma",
included_sources=["wiley/wiley-hls"],
)
for r in response.results:
print(r.title, r.url) # every result carries a citationFor builders and scientists
Developers get one endpoint, cited full-text results, and no retrieval pipeline to build or datastore to host. Scientists and clinicians get answers grounded in the peer-reviewed record — every claim opens to a source they can check.
Skip the scientific-RAG project
Rolling your own means licensing corpora, hosting a datastore, and wiring retrieval over PubMed, ChEMBL, and patents — then keeping it all fresh. Scraping gets you abstracts and hallucinated DOIs.
Valyu is one call: licensed full text, cited, commercially clean, kept fresh for you. No pipeline. No corpus licence. No six-figure commitment.
State-of-the-art across the science that matters.
DRACO is Perplexity's open expert-rubric benchmark of 100 long-form deep research tasks across 10 professional knowledge-work domains, including finance, medicine, academic research, law, and technology. Each output is graded by a per-criterion judge against a domain-expert rubric. We ran every commercially available deep research API end-to-end against the same 100 questions.
Cost vs accuracy across deep research APIs
What you can build
Drug discovery & target research
Pull mechanism, pathway, and assay literature in full, reconcile what conflicts, and return a cited brief.
Storke cut clinical evaluation timelines in half — grounding every assessment in the primary literature instead of hand-searching it.
