Amass *Core
Six specialised data layers. One retrieval engine.
One query searches literature, patents, trials, regulatory docs, and live web in parallel. Every answer cites its source — a DOI, NCT ID, or patent number. No hallucination.
Architecture
How a query fans out and comes back cited
You send one natural-language query. The engine dispatches to every relevant layer simultaneously, merges results, and returns a single JSON response with citations attached to each claim.
Data Sources
Amass *Core Engine
BioMedCore
39M+
ScholarCore
233M+
PatentCore
170M+
RegulatoryCore
FDA/EMA
TrialCore
575K+
WebCore
Live
Cited Outputs
BioMedCore
PubMed & PMC papers indexed
BioMedCore
39M+ PubMed and PMC papers — full text, MeSH-indexed
- Full PubMed and PubMed Central coverage — abstracts and full text
- Citation counts, journal impact metrics, and quality signals
- Ideal for systematic literature reviews and evidence synthesis
- MeSH-enriched metadata for precise filtering
SGLT2 inhibitors in heart failure with preserved ejection fraction
Solomon SD, McMurray JJV, et al.
N Engl J Med (2024)
ScholarCore
Scholarly papers across all disciplines
ScholarCore
233M+ papers across every discipline — OpenAlex and Semantic Scholar
- Conference papers, preprints, dissertations, and journal articles
- Cross-disciplinary coverage from OpenAlex and Semantic Scholar
- Author disambiguation and institutional affiliations
- Citation graph traversal for influence mapping
Foundation models for biomedical text mining: an empirical study
Ng, A., Bommasani, R., et al.
arXiv (2024)
PatentCore
Global patents indexed
PatentCore
170M+ patents — USPTO, EPO, WIPO — claims, families, and citations
- USPTO, EPO, and WIPO coverage via Lens.org
- Freedom-to-operate analysis and IP landscape mapping
- Claims extraction, family grouping, and citation networks
- Real-time monitoring for new filings in your domains
Bispecific antibody constructs for CDK4/6 targets
Assignee: Roche Holding AG
Filed: 2023-11-14
RegulatoryCore
FDA & EMA documents indexed
RegulatoryCore
FDA review packages, EMA EPARs, drug labels — structured for extraction
- FDA review documents, labels, and approval packages
- EMA EPARs, SmPCs, and scientific assessments
- Prescribing information and safety communications
- Structured extraction of indications, endpoints, and outcomes
Donanemab (Kisunla)
Early symptomatic Alzheimer's disease
Approved: 2024-07-02
TrialCore
Clinical trials indexed
TrialCore
575K+ trials from ClinicalTrials.gov and EU CTR — status, endpoints, sponsor
- Full ClinicalTrials.gov registry with AI-enhanced search
- Enrollment status, endpoints, and sponsor intelligence
- Phase, indication, and comparator filtering
- Trial-to-publication linking for evidence completeness
NCT04931681
Semaglutide vs placebo in NASH with fibrosis
Sponsor: Novo Nordisk
WebCore
Live web sources cited in real time
WebCore
Real-time news, press releases, and company disclosures — cited as they happen
- Real-time news, press releases, and company announcements
- Company profiles, pipeline updates, and deal intelligence
- Market intelligence and competitive landscape signals
- AI agent synthesizes and cites web sources in real time
Eli Lilly announces positive Phase 3 readout for orforglipron in obesity
investor.lilly.com
Capabilities
What you get back from every query
Cited answers
Every claim traces back to a source document. Inline citations, not hallucinated references.
Unified retrieval
Query once. Six layers respond in parallel. No need to stitch results across different databases.
Agent execution
Built-in agents plan multi-step research workflows, iterate on results, and deliver structured output.
Enterprise controls
Role-based access, audit logging, and data residency options for regulated environments.
API & MCP access
REST API for custom integrations. MCP server for Claude, ChatGPT, and Cursor — same engine, your interface.
Structured outputs
Tables, JSON, and tagged extracts. Get machine-readable results, not just paragraphs of text.
Get started
Query 500M+ documents from your terminal in five minutes
Get your first cited response in under five minutes. No prior knowledge of the data layers required.