Stop losing weeks to literature reviews.
Search across academic sources, build a structured evidence matrix, detect research gaps, and export citation-safe reviews — all in one workspace.
How it works
From search to export in six steps.
Every step produces a visible, editable artifact. No black boxes. No hidden AI decisions.
Search Papers
Query across Semantic Scholar, OpenAlex, arXiv, Exa, and Firecrawl in one search. Language-bias audit included.
Save & Screen
Select relevant papers into your project corpus. Deduplication happens automatically across sources.
Generate Matrix
AI extracts structured fields: method, dataset, key result, limitation. Edit any cell — your corrections are saved.
Explore Knowledge Map
Visualize connections between papers, methods, datasets, and limitations in an interactive force-directed graph.
Detect Gaps
Evidence-based research gap detection. Every gap references specific saved papers — no generic future work.
Export Review
Generate a citation-safe literature review. The backend validates every citation ID before export. No hallucinated references.
The problem
Literature reviews are broken. Here’s why.
0
% of AI-generated citations are fabricated
Athaluri et al., 2024
0
+ person-hours for a full systematic review
Systematic Review Guide, 2026
0
+ disconnected tools cobbled together
ResearchGold, 2026
0
years to read 105k papers from one search
LessWrong
The trust chain
Lumen breaks this chain by enforcing evidence at every step. Real papers from real sources. Structured matrix from saved papers. Gaps backed by evidence. Citations validated against your project database before export.
Core capabilities
Everything you need. Nothing you don’t.
Eight features, zero fluff. Every capability supports the evidence chain from search to export.
Multi-Source Search
Search across Semantic Scholar, OpenAlex, arXiv, Exa, and Firecrawl in one query. Partial results shown when one source fails.
Literature Matrix
AI extracts method, dataset, key result, limitation, and relevance for each paper. Edit any cell — your corrections persist.
Citation Guardrail
The backend validates every citation ID against your saved project papers. Invalid references are rejected before export. No hallucinated bibliography.
Knowledge Map
Interactive graph of papers, methods, datasets.
Research Gaps
Evidence-backed gap detection with paper references.
Conflict Detection
Flag opposing findings across studies.
PDF Ingestion
Download, extract, and chunk full-text PDFs.
Agent Audit
Full LangGraph workflow transparency.
Ready to illuminate your research?
Build defensible literature reviews with real papers, structured evidence, and validated citations.