Ananke is the intelligence layer beneath Provarion.
Ananke structures entities, relationships, evidence, exposures, scenarios, provenance, and reasoning into a shared economic intelligence layer.
Product system
Ananke
Inputs
SRCSYS: ANANKE
Intelligence layer
Product surfaces
OUTProblem and audience
Built for the Provarion platform and product family.
Financial information is scattered across tickers, portfolios, filings, news, macro series, sector classifications, models, and client conversations. Ananke exists to turn those fragments into a structured layer where relationships can be represented, tested, traced, and explained.
How it works
Ananke starts by resolving entities and economic concepts, connects them through an evidence graph, then applies exposure, scenario, relevance, provenance, evaluation, and reasoning layers before any product surface explains the result.
Key capabilities
What Ananke is designed to support.
ANANKE-01
Core architecture
Ananke combines an entity registry, economic ontology, evidence graph, ingestion layer, exposure engine, scenario engine, event relevance layer, provenance system, evaluation layer, and reasoning layer.
ANANKE-02
Entity registry
Companies, assets, sectors, macro variables, filings, news, portfolios, and client contexts are represented as inspectable entities.
ANANKE-03
Economic ontology
Economic concepts and relationships are modeled so the system can reason over exposures, dependencies, and relevance.
ANANKE-04
Evidence graph
Ananke should not create relationships because they sound plausible. A relationship exists only when supported by evidence, source metadata, method, timestamp, and confidence.
ANANKE-05
Exposure and scenario intelligence
Portfolios and entities can be connected to sectors, rates, geographies, inputs, filings, events, and macro variables so scenario context is structured before it is explained.
ANANKE-06
Reasoning with boundaries
Language models may explain and reason over structured intelligence, but they do not invent market facts, fabricate relationships, or create recommendations.
ANANKE-07
Provenance and evaluation
Outputs are designed to carry source data, method, timestamp, confidence, system version, and uncertainty.
ANANKE-08
Product power layer
Ananke powers Metis, Themis, Kairos, and Nemesis with a shared ontology, evidence graph, and reasoning standard.
Example workflow
From signal to structured context.
Boundary
Designed for decision support, not unsupported advice.
Ananke is infrastructure for analytical and workflow intelligence. It is not a trading system, brokerage, robo-advisor, or recommendation engine.
G-01
Traceable outputs
Source, method, timestamp, confidence, and uncertainty stay visible.
G-02
Structured reasoning
Models explain over structured intelligence; they do not create financial facts.
G-03
Evidence boundaries
Unsupported relationships, prices, risks, or recommendations are out of bounds.
G-04
Advisor-safe language
Communication support respects compliance-sensitive boundaries.
G-05
Uncertainty-aware
Uncertainty is shown as part of the intelligence, not hidden behind fluent text.