Interrogate the knowledge graph
Blend text similarity with structural context, audit retrieval provenance, and run forward-looking simulations without leaving the console.
Entity & edge retrieval
Tune the semantic/structural blend, flip between entity and edge views, and capture cURL-ready calls for reproducibility.
Document blending
Call POST /query/documents/hybrid to rank documents by pgvector similarity plus Neo4j structure. Each result shows the why summary and the scoring breakdown.
Provide a term above to score documents.
Explainability audit
Hit GET /explainability to inspect receipts, hybrid scores, retrieved paths, and parameter bundles for a given query.
Submit a query above to generate an audit trail.
Tier-5 reasoning search
Use POST /tier5_search for exploratory, evidence-rich answers that merge high-recall document retrieval with graph paths. Facts stay grounded while hypotheses are clearly marked.
Run Tier-5 reasoning to populate the evidence trail explorer.
Universal answerer
Use POST /query/universal_answer to fuse vector document retrieval with graph reasoning before the LLM writes the final response.
Submit a question above to run the universal answerer.
Tier-6 forensic reasoning
Use POST /tier6_reasoning to inspect evidence-backed claims, contradictions, and explicit gaps without invoking Tier-7 design mode.
Submit a misalignment query to inspect evidence coverage.
Tier-7 reasoning
Use POST /tier7/answer to run the evidence-bounded retrieval, claim extraction, hallucination check, and refusal logic pipeline.
Submit a Tier-7 query to receive a provenance-aware answer.
Reasoning trace
Use POST /query/reasoning_trace to orchestrate symbolic, neural, or hybrid graph walks and capture the chain-of-thought answer.
Provide a question above to build a reasoning trace.
Why did this change?
Compare temporal slices to surface causal hints between neighbours. Backed by GET /query/why_change.
Select an entity to analyse temporal changes.
Simulation lab
Experiment with POST /query/what_if, the multi-hop simulator, and the full scenario studio—all from a single workspace.
One-hop preview
Lightweight exploration via POST /query/what_if.
Choose a focus entity to run simulations.