Frequently asked questions
Straight answers about what Neolen does, what it doesn't do, and how to interpret what it produces.
Neolen is an AI platform that finds new therapeutic uses for drugs that already exist. Its Discovery Engine builds a biomedical knowledge graph from public science (35M+ PubMed abstracts, drug databases, gene/protein interaction data, and clinical trial registries) and reasons over it to generate ranked drug-repurposing hypotheses. Its Evidence Engine then checks those hypotheses against real hospital data using natural-language cohort discovery.
Drug repurposing (or repositioning) is finding a new disease indication for a drug already approved for a different use. Because the drug has already cleared human safety testing, repurposing candidates can, under the FDA's 505(b)(2) pathway, sometimes skip extensive preclinical work and Phase I entirely and move straight to Phase II efficacy — a fraction of the $1–2.6B and 10–15 years a novel drug typically requires.
General-purpose language models can summarize what's already been said about a drug, but they don't systematically traverse a structured, versioned knowledge graph of drug, gene, protein, pathway, and disease relationships, apply link-prediction and mechanistic path-reasoning models, filter out candidates that already failed in ClinicalTrials.gov or show adverse-event red flags in FAERS, or render a full explainable evidence path back to source literature. Neolen is purpose-built for this task, not a general chat interface.
No. Neolen explicitly does not design novel molecules — that is generative chemistry and structural biology territory (the domain of tools like AlphaFold). Neolen focuses on finding new indications for drugs that already exist and have already been through human safety testing.
No. Neolen does not run wet-lab or in-vivo validation, and it does not run clinical trials. It generates and ranks hypotheses from public data, then checks them against existing real-world patient data where available. A human researcher or clinician decides what happens next.
No. Neolen outputs are labeled as hypothesis-supporting signal, never as proof of efficacy. Observational data and literature-based inference generate leads; only controlled clinical trials establish causation. This distinction is treated as a hard product requirement, not a footnote.
The platform is split by a hard architectural boundary. The Discovery Engine touches only public data and runs as multi-tenant SaaS. The Evidence Engine touches patient data and is deployed inside the customer's own VPC or on-premise environment — hypotheses flow in, and only aggregate, de-identified statistics ever flow out. Reach out to us for details on our security and compliance posture.
No. The Discovery Engine is designed to deliver value entirely on public data from day one, with no hospital data partnership, BAA, or HIPAA audit required to get started. The Evidence Engine, which does use real-world patient data, is a later-phase capability deployed only inside a customer's own data environment.
Fourteen priority public sources at launch, including PubMed/MEDLINE, SemMedDB, DrugBank, ChEMBL, UniProt, Reactome, DisGeNET, CTD, STRING, HPO, PharmGKB, ClinicalTrials.gov, OpenFDA FAERS, and UMLS as the normalization backbone.
Every hypothesis ships with its full evidence path back to source papers and database entries, plus separate scores for mechanistic plausibility, evidence strength, novelty, safety, feasibility, and regulatory viability. As a product-level credibility test, the system must pass a blinded retrospective validation: restricted to pre-2019 data, it must independently rediscover baricitinib for COVID-19 and 15–25 other known repurposing successes.
Four groups: translational researchers and academic PIs generating publishable hypotheses; pharma R&D and portfolio strategists de-risking 505(b)(2) candidates for in-portfolio assets; rare disease foundations seeking any plausible existing treatment; and biotech founders sourcing undervalued in-licensing opportunities.
Neolen is recruiting 2–3 rare disease foundations and one academic PI as founding design partners, with free or near-free access in exchange for structured feedback and a published case study. Reach out via our contact page to apply.
Every Cure, BenevolentAI, Insilico Medicine, and Recursion are all active in adjacent parts of this space, and the field is well funded and increasingly competitive, particularly in rare disease. Neolen's differentiation is the closed loop: generating a hypothesis from public data and then validating it against real-world patient evidence in the same system, which we believe is a less crowded and more defensible half of the market than hypothesis generation alone.