Research Startup Radar

Claude Science emphasizes full provenance tracking on every result to enable traceability and reproducibility, directly tackling a core pain point for...

Clemson Libraries' decision to add Web of Science AI Research Assistant for campus-wide use signals rising institutional demand for AI tools embedded...

Anthropic's new Claude Science workbench and NVIDIA's BioNeMo Agent Toolkit integrate literature review, code execution, and domain-specific models...

Semantic Scholar provides the free data backbone powering tools like Elicit and ResearchRabbit, while SciSpace targets end-to-end workflows with...

Elsevier's LeapSpace expansion adds Writing Coach and Claim Radar agents that draft manuscripts and verify claims against peer-reviewed literature...

Literature Review Tool Benchmarks

• 🔥 7 Tools Compared: AI for Literature Review 2026: 7 Tools Compared singles out Elicit as the strongest...

The AI literature review space is rapidly specializing across search, synthesis, gap detection, and pre-submission review.

• Scite.ai is evaluated...

A team built a persistent agent memory layer on Elasticsearch achieving 0.89 recall. This infrastructure approach could inform founders designing long-term memory for research agents.

The CAT Co-Intelligence Framework offers a practical design blueprint for AI research assistants that preserve critical thinking.

• Core loop: Think...

Scite stands out as one of the strongest tools for mapping how later literature cites and evaluates earlier papers, delivering citation-context signals that founders can study when designing verification and evidence-based discovery features.

Digital Science's Dimensions Citation Check API gives journal editors an automated way to detect and assess citation manipulation, strengthening verification tools for research integrity.

Sakana AI's first commercial product, Sakana Marlin, is positioned as a business-focused autonomous research assistant. This move highlights early...

Big Tech AI Tool Launches

• 🔥 Microsoft Discovery Platform: Microsoft Discovery Platform brings agentic AI to scientific work where AI agents...

Cited Evidence Chat with PDF lets researchers upload papers and query key findings, data tables, and variables instantly while verifying answers against the original text, turning hours of skimming into seconds.

• Google tools focus on literature review and data analysis for researchers
• Microsoft Discovery uses graph-based engine to connect proprietary and external data
• Divergent paths: task-specific assistance vs agentic knowledge integration

Infrastructure Barriers

• 🔥 Full-Text Access Barrier: Restricted full-text access and scholarly paywalls are identified as the primary barriers...

ScholarLM V5's new YOLO Mode runs a fully autonomous literature review loop, from planning to cited synthesis, with zero manual steps.

• Core loop:...

Full-text access and scholarly paywalls are the primary barriers to developing accurate AI tools for research papers. Founders building discovery and summarization products must treat access solutions as a foundational priority.

Missing metadata means papers may not even appear to exist in research systems, highlighting the urgent need for better provenance infrastructure to fix leaky pipelines.

A new paper proposes an AI-powered multi-modal research paper assistant designed to streamline workflows through semantic search and automation.