DataFab Platform Documentation

Welcome to the DataFab platform architecture documentation.


Overview

DataFab is an AI-powered data intelligence platform built on a metadata-driven architecture that provides unified access to distributed data assets while maintaining strict security controls. The platform enables natural language interaction across all components, supports configurable automation levels, and integrates with 200+ data sources through its Knowledge Fabric, Studio, Dialog, Exchange, and Graph RAG modules. The platform’s primitives are used to build domain-specific Utilities (Transaction Monitoring, Car Finance, Compliance) — packaged solutions that compose the same components into end-to-end applications for a particular operational use case.


Documentation

Document Description Version
Introduction Platform overview, capabilities, and security summary 3.5
Architecture System architecture, deployment models, security boundaries 5.2
Knowledge Fabric Knowledge graph, entity resolution, MCP connectors, data integration 4.0
Studio DDAs, widgets, datasets, utilities, Chain of Agents, MCP integrations 6.2
AI & LLM LLM security, output consistency, model provenance 3.0
CI/CD CI/CD pipeline security 2.0
Security Operations SOC, monitoring, incident response 2.0
Graph Operations Rule Engine, DAG Rules, graph workflows 4.3
Schema Management Business domain discovery, schema registry 2.0
Compliance Capabilities Platform compliance features, data hold, retention 3.0
API Security API gateway, authentication, rate limiting 5.0
Exchange Data asset marketplace, catalog, wallet & blockchain, metering, access 4.0
Graph RAG Graph-enhanced retrieval, hybrid search, Text-to-Cypher Rules 3.1
Trust & Compliance Managed deployment security, compliance, and governance 2.0
Dialog Conversational layer — playbooks, context-awareness, self-awareness, widget rendering 2.3

Utilities

Utility Description Version
Utilities — Overview Utility framework, anatomy, catalog, lifecycle, security posture 1.0
Transaction Monitoring Customisable DAG chain, OSINT + screening, BPMN per tenant, explainability + HITL, SAR pipeline, STP, dashboards 2.2
Car Finance (AutoFab) Motor finance remediation under the FCA Motor Finance Scheme — eight-stage lifecycle 1.0
Compliance Conflict-of-interest detection, OSINT schemas, BPM-driven processes, risk rules, watchlist screening 1.0

Platform Components

Knowledge Fabric

The foundational data integration layer with 200+ MCP connectors, persistent knowledge graph, entity resolution, and enterprise data integration across structured, semi-structured, and unstructured sources.

Studio

Development environment for building Data-Driven Agents (DDAs), widgets, datasets, utilities, and multi-agent workflows (Chain of Agents, with Graph of Agents planned). Features domain-driven creation flows, MCP integrations, AI hybrid planning, semantic asset search, and five configurable operational modes.

Dialog

The platform’s conversational layer — the analyst-facing surface that turns the rest of DataFab into something a person can talk to. Dialog is a peer to Studio, Knowledge Fabric, Graph Operations, and the AI/LLM Layer; it owns playbook-based communication, context-awareness, system self-awareness, and inline rendering of Studio Widgets.

Exchange

Data asset marketplace enabling organizations to publish, discover, acquire, and monetize data assets. Features blockchain-backed transactions (DAAC token on Ethereum), usage-based metering, tiered access plans, and a double-entry ledger for transparent revenue sharing.

Graph RAG

Graph-enhanced retrieval augmented generation combining vector similarity search with knowledge graph traversal for contextually rich, relationship-aware responses. Includes Text-to-Cypher Rules — natural-language graph-querying with schema-bounded generation and deterministic fallback.

Graph Operations

Operational intelligence module providing graph-based pattern detection, rule engines, DAG Rules (self-organising rule chains), workflow automation, entity assessment, and anomaly scoring across connected data.

AI & LLM Layer

Provider-agnostic LLM integration via the LLM Router with output consistency controls, model provenance tracking, and quality assurance.


Utilities

DataFab Utilities are packaged, domain-specific applications built on the core framework. Each utility composes the same primitives — Knowledge Fabric, Studio, Dialog, Schema Management, DAG Rules, Text-to-Cypher, BPMN workflows, AI/LLM Layer — into an end-to-end solution for a particular operational use case. Utilities reuse platform security, audit, and operational-mode controls; they do not introduce parallel infrastructure.

Transaction Monitoring

Financial-crime alert triage and case-management application. Customisable DAG rule chain (69 platform rules plus customer-authored rules across 12 rule classes and 10 logic patterns), multi-source OSINT and watchlist screening, customer profile and behavioural intelligence, anomaly detection, evidence assembly, linked-accounts and flow-of-funds analysis, tenant BPMN workflow, explainability with Human-in-the-Loop, SAR filing pipeline (FinCEN / NCA), outreach gap detection, straight-through processing for clear cases, dashboards and analytics, policy versioning with ground-truth testing, and a full analyst workbench (graph renderer, visual link analysis, evidence panel, typologies drawer, baseline + spike chart, raw-transactions table, recommendations cards, audit-trail timeline).

Car Finance (AutoFab)

Motor finance remediation utility supporting end-to-end execution under the FCA Motor Finance Scheme. Eight-stage lifecycle from data ingestion through population build, cohorting, scheme testing, redress calculation, customer outreach, payment validation, and case closure with complete audit trail. Reference deployment: AutoFab — PwC private cloud instance for Conister Bank.

Compliance

Compliance and conflict-of-interest utility for legal-tech and regulated-industry use. Six-capability surface: graph-based conflict detection, OSINT schema management, BPM-defined compliance processes, risk rules with versioned policies, adverse-media checks, and watchlist screening with built-in catalogue plus tenant-selectable lists.



Security Highlights

Principle Implementation
Defense in Depth Multiple independent security layers
Zero Trust All requests authenticated regardless of source
Encryption Everywhere AES-256 at rest, TLS 1.3 in transit
Schema-Driven Security All extraction bounded by user-defined schemas
Human-in-the-Loop Configurable automation with escalation workflows

Regulatory Compliance

The platform supports compliance with GDPR, CCPA/CPRA, HIPAA, PCI DSS, SOC 2, EU AI Act, and industry-specific standards.


Contact

For security inquiries or additional documentation, please contact the DataFab security team through your account representative.