FieldGuide
A research-grade reading companion that ingests a stack of papers in a field and auto-builds a knowledge graph of concepts, methods, and how they relate.
PhD students and postdocs orienting themselves in a new research field
- Drop in 50 PDFs, get a navigable concept map with citations to the originating paper
- Daily "today's three new connections" digest as more papers are added
- Export the graph as a slide-ready visual for a literature-review section
- Track which claims are well-supported across many papers vs. only one
OntoLearner is the first open-source framework that benchmarks ontology learning across domains — a non-engineer academic can now stand on real infrastructure instead of bespoke scripts.
Real, persistent pain: Paperguide's research shows researchers spend 15–20% of research time on literature and 4 hrs/week searching; Elicit claims 2M+ researchers using AI reading/extraction tools, confirming a large addressable base for PhD/postdoc literature onboarding.9 Best AI Research Assistant Tools in 2026 (Free + Paid) ↗Elicit Pricing ↗
Crowded — Paperguide, Elicit (2M users, 125M+ papers), SciSpace (270M papers), Consensus, ResearchRabbit, Connected Papers, NotebookLM, Undermind, and Semantic Scholar all already serve literature discovery, extraction, and relationship/graph views; no open white space for a 'knowledge graph from papers' wedge alone.9 Best AI Research Assistant Tools in 2026 (Free + Paid) ↗Undermind: your AI co-researcher for the literature ↗
Weak: paid tiers sit at $10–$49/mo (Consensus $10, Paperguide $12, Elicit $49) with generous free plans and most academic users on institutional/budget-constrained grants; selling to PhDs/postdocs specifically is the lowest-converting academic segment.9 Best AI Research Assistant Tools in 2026 (Free + Paid) ↗Elicit Pricing ↗
Literature-orientation pain is structural and persists, and OntoLearner (real, on PyPI, arXiv 2607.01977) gives a defensible technical foundation; however, the application layer above ontology-learning infra is commoditizing fast as every incumbent adds graph/synthesis features.OntoLearner · PyPI ↗OntoLearner: A Modular Python Library for Ontology Learning (arXiv) ↗
High — OntoLearner is published, pip-installable, and explicitly unifies ontology access + LLM learning + benchmarking across domains, so the hard ontology-learning work is already done; remaining work is PDF ingestion, graph storage, and a concept-graph UI, all well-trodden engineering.OntoLearner · PyPI ↗OntoLearner GitHub ↗