Documentation / Concepts
Signal Provenance
Signal provenance preserves where information originated, how it moved, and how it was used in decision reasoning.
Provable decision systems require more than accurate outputs. They require the ability to inspect the signals that shaped those outputs, where those signals originated, and how they were interpreted over time.
What is Signal Provenance
Signal provenance is the record of a signal's origin, lineage and contextual path through the decision environment.
It answers a simple but critical question: what is this signal, where did it come from, and how did it influence the resulting decision?
Why Provenance Is Lost
In conventional data pipelines, provenance often becomes difficult to preserve once signals are joined, aggregated, exported or transformed into downstream AI workflows.
- signals are merged into flat analytics structures
- inference is mixed with asserted fact without distinction
- copied outputs lose the source path that created them
- decision systems retain results but not the reasoning chain beneath them
The Congregation DB Approach
Congregation DB is designed to preserve origin and reasoning context as structural properties of the system rather than as optional metadata appended after the fact.
Source attachment
Signals retain origin details so downstream systems can still inspect where information entered the environment.
Truth lane separation
Deterministic, probabilistic and derived signals remain legible rather than being collapsed into a single layer.
Contextual graph reasoning
Relationships, dependencies and surrounding context remain part of the signal trace as reasoning progresses.
Conceptual Flow
Provenance flow
Signal origin remains attached as information moves from source capture into contextual reasoning and decision output.
Provenance path
A signal enters from an operational system, API, document or model output.
The signal is preserved in the lane that reflects its type and certainty.
The signal participates in context synthesis without losing origin or lineage.
The outcome remains tied back to the signal path that informed it.
As signals move through the system, origin and contextual lineage remain attached instead of being flattened away.
Provenance in Reasoning
Provenance
Reasoning remains auditable only when the signals beneath a decision still expose origin, type and context.
Provenance is not only about recording source systems. It also preserves the reasoning environment around a signal so downstream systems can inspect:
- where information originated
- what class of signal it represented
- what relationships shaped the reasoning path
- how the resulting conclusion was formed
Example Signal Trace
The following conceptual trace shows how a signal can remain legible from origin through contextual reasoning into a decision output.
signal_id: supplier_default_notice source_system: external_compliance_api captured_at: 2026-02-04T10:14:00Z truth_lane: deterministic related_entity: portfolio_company:arcadia_energy graph_context: - supplier: helix_grid_components - dependency: battery_supply_chain used_in: exposure_risk_assessment provenance: retained
decision_output: exposure_risk = elevated
derived_from:
- signal: supplier_default_notice
truth_lane: deterministic
source: external_compliance_api
- signal: logistics_disruption_probability
truth_lane: probabilistic
source: market_model_output
context_path:
portfolio_company -> supplier -> shipping_route
traceability: inspectableDesign Principle
Design principle
Signals should never lose origin, type or contextual lineage as they move toward a decision.
Why This Matters
As AI systems move from answering questions toward shaping real-world decisions, provenance becomes part of the infrastructure requirement rather than an optional reporting feature.
Systems that cannot preserve signal origin may still produce outputs, but they struggle to support the level of traceability, auditability and confidence demanded by serious decision environments.