Idea Network × Idea Interflowing · Live
Infrastructure for agents, an inspectable workspace for people
Idea Network organizes papers, authors, concepts and citation links into a research graph agents can call directly. Idea Interflowing turns that infrastructure into an inspectable workspace. It reports what it found while explicitly labeling not-found evidence, uncertainty and graph blind spots.
How It Works / Three core capabilities
General models excel at retrieving and summarizing within the known. Idea Interflowing works on the structure of the citation network itself — where the gaps are, where the frontier is heading, and how cross-disciplinary connections hold up.
Locate the "voids" in the citation network — the structural gaps surrounded by dense citation yet truly reached by no one. Agents then work out which not-yet-posed research question each void corresponds to.
Identify a field's current research frontier, trace its advance along the time axis, and answer the hardest and most valuable question of all: where is the next breakthrough most likely to appear?
Find inspiration along real academic relationships and generate candidate hypotheses where fields intersect. Not free association, but reasoning along the real structure of the citation network — every hypothesis carries an evidence chain that traces back to the source.
Live Workflows
Idea Creator turns a direction into a structured proposal. Idea Supervisor grounds review judgments in inspectable evidence paths. Both workflows are developed and live.
Start from a research direction and follow connected papers, concepts and citations to find structural gaps worth pursuing. Separate established evidence, refutable hypotheses and open validation work, then deliver a structured proposal ready for review and execution.
Review a paper or abstract against target-venue dimensions and expose the evidence path behind every judgment. Explicitly separate graph-confirmed evidence, graph-not-found results and graph blind spots so each score can be challenged and refined.
FAQ / Frequently asked
Idea Network is agent-native research-graph infrastructure for papers, authors, concepts, citation tools and stateful sessions. Idea Interflowing is the graph-grounded workspace above it, turning those capabilities into workflows researchers can inspect, challenge and continue editing.
Ordinary literature tools help you "read faster" — they answer "what already exists in this field." Idea Interflowing is infrastructure — it turns the whole citation network into an object any LLM can freely roam and reason over, letting an agent autonomously answer "what is still missing, and where the next breakthrough lies," with every conclusion carrying an evidence chain you can click back to the source.
Idea Creator turns a direction and connected evidence into a structured proposal. Idea Supervisor reviews a paper or abstract against target-venue dimensions and exposes evidence paths, not-found results and graph blind spots. Both workflows are live.
Yes. Idea Network has a public preview at wentor.ai/network. Idea Interflowing is live at interflowing.ai and continues to iterate in beta.
Every citation comes from authoritative academic data sources and can be traced one by one. Through our in-house deep-research framework and knowledge graph, we keep both the citation and the claim it supports faithful, ruling out fabricated references and misattributions by design.
Both products are live: let agents find evidence in Idea Network, then turn judgments into inspectable deliverables in Interflowing.
Product Feedback / Co-create
The product is live. Share a real research problem and your contact details to help us calibrate the workflow.
Glue literature, code, experiments, and writing into a single local research operating system.
Learn more → ThesisAgentAI academic workbenchFrom topic to finished manuscript — a human–AI co-creation workbench for verifiable academic writing.
Learn more → HashMindAI research communityA knowledge and asset layer built for agents — the StackOverflow of the agent world.
Learn more →