Turn a question into a fully-recorded investigation — sources ranked, hypotheses weighed, findings mapped onto a living graph of fields and concepts.
Record types
The structured data this adds to your pod.
Question
An open question driving an investigation — the thing research answers.
- Status
Research
An investigation — the sources consulted, the findings, the confidence, and the conclusion.
- Status
- Conclusion
- Confidence
Field
A domain of knowledge — Mathematics, Physics, AI, Philosophy. Fields overlap and feed each other.
- Description
- Also known as
Concept
An atomic idea — Bayesian inference, gradient descent, qualia. Belongs to one or more fields; has prerequisites.
- Definition
- Also known as
Source
Anything consulted — a paper, article, dataset, or post. Ranked by credibility and NEVER discarded; rejected sources are kept with a reason.
- URL
- Author
- Published
- Type
- Credibility
- Quality (0–100)
- Status
- Why rejected
- +1
Hypothesis
A candidate theory or answer under test — argued for and against by sources.
- Statement
- Stance
- Confidence
- Reasoning
Finding
A validated fact — the thing established between the question, the sources, and the reasoning. Carries calibrated confidence and links to its evidence.
- Claim
- Confidence (0–1)
- Evidence summary
- Status
Views
- Knowledge MapGraph
- Open QuestionsList
- SourcesTable
- HypothesesBoard
- FindingsTable
- Research LogTable
Playbooks
Repeatable runs your agent can start for this project.
- Research a QuestionInvestigate a question end-to-end — gather and rank sources (keeping the rejected ones), weigh hypotheses, emit findings with calibrated confidence, and map them onto the knowledg…
Relationships
- Field FieldOverlaps with
- Concept FieldPart of
- Concept ConceptPrerequisite of
- Concept ConceptRelates to
- Research QuestionInvestigates
- Hypothesis QuestionAnswers
- Research SourceExplored
- Research FindingConcluded
- Source HypothesisSupports
- Source HypothesisContradicts
- Source SourceCites
- Finding ResearchDerived from
- Finding SourceSupported by
- Finding ConceptAbout
First conversation
After install, your agent sets the project up with you. It starts here:
Seed the workspace with the first real questions worth researching and the fields they live in — so the agent has something concrete to investigate, not a blank graph.
- “What's a question you actually want a well-evidenced answer to — one where a good answer would change what you do?”
- “Which field(s) does it sit in — and is there a neighbouring field it might borrow from?”
- “What would count as a convincing answer, and what would change your mind?”