GQ
Greyquill · Governed AI for regulated enterprises

Greyquill Platform, C4 Architecture

Compliant AI · Trusted Data. One platform with a stable governed core and a deep customization layer on top, deployed in your environment, with the audit trail as a first-class outcome. "How did this number get here?" is one query, not three weeks.

Built on IBM, owned by you GQ Data · trusted data & lineage GQ Govern · diagnose, policy & evidence GQ Agents · governed execution GQ Studio · the platform UI shell + customization Vertical Packs · Insurance · BFSI
C4 · Level 1

System Context

Audience: CEO / Board · CISO / Procurement

Where Greyquill fits in the enterprise landscape, who uses it, what feeds it, and who consumes its governed output, without exposing internals.

Users

CDO / Data Steward

Manages trusted master data & resolution.

Compliance Officer

Needs evidence, lineage & audit answers.

AI Engineer

Builds on governed, AI-ready context.

Greyquill Platform · Governed AI Foundation

Greyquill Platform Stable governed core + customization layer · in your OpenShift cluster
GQ Data

Master data, resolution, lineage, quality.

GQ Govern

Diagnose, policy, controls, evidence.

GQ Agents

Governed workflow execution + runtime evidence.

GQ Studio

The one platform UI: workspaces + every layer's surfaces + low/pro-code customization.

Vertical Packs

Pre-built domain models & rules: Insurance (first), BFSI. The "velocity pack" go-to-market.

Consumers

BI & Analytics

Tableau, Power BI, Looker via JDBC/ODBC.

Audit / GRC

ServiceNow, RSA Archer. Signed audit packs.

Regulators

EU AI Act, GDPR, fair-lending evidence.

Data Sources (in) & Unstructured Input

SAP / Oracle ERP

Core Banking

Salesforce CRM

S3 / Blob / GCS

Kafka / Streams

PDFs · Scans · Email

C4 · Level 2

Containers on Red Hat OpenShift

Audience: CTO / Enterprise Architect · CISO

The applications and data stores inside the OpenShift boundary, and the contract-only interfaces between layers (no shared code, no shared database).

⬡ Red Hat OpenShift 4.x · namespace-per-tenant isolation
GQ Data

Governed data foundation

GQ Ingest Connectors · batch + CDC · quarantine
GQ Resolve Match · merge · golden records
GQ Classify PII · financial · custom tags
GQ Lineage Provenance · temporal versioning
GQ Quality 6-dimension scoring · drift
GQ Graph API REST + GraphQL · OAuth2/RBAC
GQ Govern

Diagnose · compliance · risk

Assess AI-maturity & readiness diagnosis
Policy Authoring · push to GQ Data
Inventory Model & AI-asset register
Diagnose Unstructured document intelligence
Evidence Factsheets · audit assembly
GQ Agents

Governed activation

Agent Runtime Prebuilt + custom agents
Evidence Stream Real-time runtime audit capture
Human Oversight Approve · hold · override (logged)
GQ Studio: The Platform Shell  (one app; customization is a section inside it, the "Salesforce move")

Every layer's UI lives here: sections owned by GQ Data / GQ Govern / GQ Agents, plus the clicks → code spectrum

App Shell & Workspaces Role-based, residency-bound governed environments
Workflow Canvas One composer, block families from Data · Govern · Agents
No-code Config Validation · dashboards · views
Low-code Extend Custom entities · connectors
Pro-code SDK Custom agents · frameworks
Vertical Packs

Pre-built domain models on the core ontology

Insurance Pack First: claims, KYC, policy
BFSI Pack Accounts, loans, AML, cards
+ future verticals Telecom · Retail · Health

Shared Data Stores

Neo4j

Entity graph · lineage

PostgreSQL

Relational index · scores

MongoDB

Raw records · packs

Kafka

Event bus

Redis

Cache · policy state

Elasticsearch

Search · fuzzy match

Key Interfaces, contract-only

GQ Data → GQ GovernEventsquality.violation · lineage.updated · ingestion.completed
GQ Govern → GQ DataPoliciespolicy.push · policy.deactivate · assessment.request
GQ Data → GQ AgentsGoverned ContextGraphQL / REST entity + lineage queries
GQ Agents → GQ DataEvidence Streamruntime audit events for pack assembly
GQ Govern → GQ AgentsAgent Policyguardrail directives for execution
GQ Studio → all layersUI, config & extensionsone shell over the layers' public APIs; declarative config + SDK
C4 · Level 3

Components: GQ Data Internal Flow

Audience: Engineering · Sprint planning (Phase 1 build)

The internal flow of GQ Data, the first product being built (Phase 1: GQ Ingest + GQ Resolve). Raw records become governed, queryable context.

INGEST
Connector Engine
Schema Discovery
Quarantine Manager

→ MongoDB (raw records)

RESOLVE
Match Engine
Merge Engine
Golden Record Store

→ Neo4j + PostgreSQL

CLASSIFY
PII Detector
Financial Classifier
Custom Taxonomy

→ PostgreSQL (tags)

QUALITY
Rule Engine
Score Aggregator
Drift Detector

→ PostgreSQL + Kafka

SERVE
REST Gateway
GraphQL Engine
Audit Pack Assembler

reads Neo4j + PostgreSQL + MongoDB

Flow: Raw Records → Golden Records → Tagged Entities → Scored Entities → Governed Context API  ·  boundaries: external systems feed in (top) · GQ Govern (policy push / events out) & GQ Agents (context API / evidence in) on the right.