LIVE COMPARISON

Same model. Different substrate.

Same question. Same Claude Sonnet.

First: Claude with no graph context.
Then: ForcedAlpha + Claude, grounded in 3,900+ nodes, 12,800+ edges, and live cascade analysis.

Your question → planner interprets intent → graph lookup → nodes + paths injected → Sonnet writes answer
The graph answer takes longer because it plans, queries, and traverses real dependency paths before generating.
CLAUDE SONNET No graph context
Generating ...
FORCEDALPHA + SONNET Graph-augmented intelligence
Searching graph ...
Edge provenance

How every relationship is backed

47% carry a clickable source URL. Another 39% are tagged with a documented provenance reason — industry-structural, internal model, transitive, regulatory, paywalled. The remaining 4% are queued for re-mining or pre-taxonomy uncategorized.

  • Direct source URL 54.8%
  • Industry knowledge 18.8%
  • Internal model 5.2%
  • Transitive 4.5%
  • Pending 5.9%
  • Uncategorized (legacy) 7.0%
  • Other tagged 3.8%

47% of edges carry a clickable source URL. Another 39% are tagged by provenance type — industry-structural, internal model, transitive, regulatory, paywalled. The remaining 4% are queued for re-mining or pre-taxonomy uncategorized.

ForcedAlpha · [email protected]