Confidence Tier
An evidence grading system (VERIFIED/INFERRED/ASSUMED) that tags research claims by reliability, preventing hallucinated data from reaching deliverables.
Confidence tiers enforce intellectual honesty in Knowledge OS research output. VERIFIED means direct evidence found and cited (a specific document, URL, or data point). INFERRED means deduced from combining 2+ pieces of evidence with stated logic. ASSUMED means the claim is plausible but lacks direct support. Every research skill (/research-prospect, /hypothesis-builder, /meeting-prep) tags claims with tiers. Downstream skills treat tiers differently: VERIFIED claims go into deliverables as-is, INFERRED claims get hedging language, ASSUMED claims trigger re-research or get dropped. The system was built after a fabricated prior-issue citation in a consulting deliverable.
Where it shows up:
Related Terms
Architecture
Upstream/Downstream
The separation of research agents (describe reality) from recommendation agents (interpret and activ...
Methodology
ICP Model
An Ideal Customer Profile definition with firmographic criteria, behavioral signals, and pain hypoth...
Methodology
Quality Gate
An automated checkpoint in a skill chain that evaluates output against defined criteria before allow...