METHOD Context grounding
→ SEARCH Evidence & anomaly research
→ LLM Hypothesis exploration
→ VALIDATE Counter-evidence & trust
→ GOVERN Recommendation / next step
AI 01
Feed & Source Quality
Source ingestion quality, completeness, exceptions and trust.
Records — Evidence — Sources — Trust UNASSESSED
CONTEXTUAL SUMMARY Source ingestion quality, completeness, exceptions and trust. Exact source evidence will be loaded for this investigation.
RECOMMENDATION Review the weakest source-quality evidence before changing ingestion or posting controls.
AI RESEARCH QUESTION Where is feed quality, completeness or provenance weakest?
AI 02
Posting Anomalies & Throughput
Posting delay, backlog, throughput and anomaly signals.
Records — Evidence — Sources — Trust UNASSESSED
CONTEXTUAL SUMMARY Posting delay, backlog, throughput and anomaly signals. Exact source evidence will be loaded for this investigation.
RECOMMENDATION Investigate material posting delays and anomaly clusters before period-end processing.
AI RESEARCH QUESTION Which posting delays or throughput changes are material and evidence-supported?
AI 03
Classification Drift
Classification confidence, mapping drift and category integrity.
Records — Evidence — Sources — Trust UNASSESSED
CONTEXTUAL SUMMARY Classification confidence, mapping drift and category integrity. Exact source evidence will be loaded for this investigation.
RECOMMENDATION Review low-confidence or drifting classifications through governed mapping controls.
AI RESEARCH QUESTION Where is classification confidence drifting from the established pattern?
AI 04
Exception Concentration
Exception concentration by source, entity and returned population.
Records — Evidence — Sources — Trust UNASSESSED
CONTEXTUAL SUMMARY Exception concentration by source, entity and returned population. Exact source evidence will be loaded for this investigation.
RECOMMENDATION Prioritise the highest evidence-supported concentration before remediation.
AI RESEARCH QUESTION Which sources or entities explain the strongest supported exception concentration?
AI 05
Account Behaviour
Account movement and behavioural patterns supported by returned ledger data.
Records — Evidence — Sources — Trust UNASSESSED
CONTEXTUAL SUMMARY Account movement and behavioural patterns supported by returned ledger data. Exact source evidence will be loaded for this investigation.
RECOMMENDATION Trace unusual account behaviour to exact transactions before changing account treatment.
AI RESEARCH QUESTION Which account movements differ materially from the established behaviour?
AI 06
Unusual Transaction Patterns
Anomalous transaction patterns supported by feed and matching evidence.
Records — Evidence — Sources — Trust UNASSESSED
CONTEXTUAL SUMMARY Anomalous transaction patterns supported by feed and matching evidence. Exact source evidence will be loaded for this investigation.
RECOMMENDATION Review exact anomalous records and supporting evidence before escalation.
AI RESEARCH QUESTION Which transaction patterns are unusual, material and supported by exact records?
AI 07
Evidence & Data Trust
Evidence completeness, provenance, lineage and trust signals.
Records — Evidence — Sources — Trust UNASSESSED
CONTEXTUAL SUMMARY Evidence completeness, provenance, lineage and trust signals. Exact source evidence will be loaded for this investigation.
RECOMMENDATION Request missing evidence or lineage before relying on weakly supported conclusions.
AI RESEARCH QUESTION Where is evidence incomplete, stale, weakly linked or low trust?
AI 08
Reconciliation & Control Integrity
Matching, audit and control-integrity evidence relevant to the Ledger Feed.
Records — Evidence — Sources — Trust UNASSESSED
CONTEXTUAL SUMMARY Matching, audit and control-integrity evidence relevant to the Ledger Feed. Exact source evidence will be loaded for this investigation.
RECOMMENDATION Resolve material control or reconciliation differences through governed review.
AI RESEARCH QUESTION Which reconciliation or control-integrity signals require governed investigation?