Lune launches scientific search for AI agents

Lune’s strongest idea is treating scientific research as an agent toolchain, not another standalone chatbot. Narrowing the corpus to high-quality venues should improve trust, though it also makes coverage less comprehensive than broad academic search engines.

• –MCP integration makes grounded research available inside developers’ existing AI workflows.
• –Full-text retrieval, citation tracing, and claim checking give agents better provenance than relying on model memory.
• –Guidance for literature reviews, evaluations, ablations, and peer review extends the product beyond paper discovery.
• –The curated corpus is a meaningful differentiator, but users must accept narrower coverage than Semantic Scholar-style indexes.
• –Evidence retrieval reduces hallucination risk; it cannot eliminate weak synthesis or flawed interpretation by the agent.