CRR — Regulation (EU) No 575/2013 — Article 92

Institutions shall at all times satisfy the following own funds requirements: a Common Equity Tier 1 capital ratio of 4.5%, a Tier 1 capital ratio of 6%, and a total capital ratio of 8% of the total risk exposure amount calculated in accordance with Article 92(3)CRR Art. 92(3) → total risk exposure calculation and Article 92(4)CRR Art. 92(4) → additional own funds requirements.

Extracted logic
CET1 ratio = CET1 / REA ≥ 4.5%
T1 ratio = T1 / REA ≥ 6%
TC ratio = TC / REA ≥ 8%

Regulation you can
finally see through

The AI operating system that transforms dense regulatory text into structured knowledge your bank can act on.

Request a Demo See How It Works
2.7M knowledge nodes 90%+ accuracy 21 frameworks

The platform

Like Claude Code
for banking

An AI agent with access to your bank's regulatory knowledge, data model, and building tools. You ask questions. It researches with expert precision. You stay in the loop.

Four-phase deep research pipeline with parallel AI workers
Interactive evidence graphs with full citation traceability
Build executable models directly from research findings
Request a Demo
Workspaces
Research
Institution Context
Build
Production
Knowledge Base
CRR / CRD
EBA Guidelines
SFSA / FFFS
Chat
Evidence Board
Report
Sources
What are the capital requirements for our mortgage portfolio under CRR Article 125?
Based on CRR Article 125, exposures fully secured by mortgages on residential property shall be assigned a risk weight of 35%, subject to conditions in Article 125(2). CRR II (Regulation 2019/876) amends this with an exposure-to-value dependent approach…
CRR Art. 125(1)CRR II Art. 1(42)EBA/GL/2020/06

Three pillars of knowing

Same regulation.
Three layers of understanding.

Every regulation BARK processes is decomposed into structural text, semantic concepts, and computational logic.

PILLAR I

Structural — Legal Text

Full legal text parsed into a traversable hierarchy: Document → Part → Title → Chapter → Article → Paragraph → Point. Cross-references modeled as graph edges.
5,247
Articles
80K+
Elements
PILLAR II

Conceptual — Semantic Knowledge

18,343 banking concepts extracted and matched to the official EBA Data Point Model vocabulary with IS_A hierarchies and semantic embeddings.
18,343
Concepts
DPM
Matched
PILLAR III

Computational — Executable Logic

4,786 formalized rules with parameters and mathematical functions. Requirements extracted from prose into executable logic, each traceable to its source article.
4,786
Rules
568
Functions
2,700,000 nodes. One graph. Full traceability.

Capabilities

Purpose-built for
regulatory intelligence

Not another chatbot. A structured knowledge system with verification, traceability, and computational logic built in.

Regulatory Knowledge Graph

2.7 million nodes across three pillars. Not chunks in a vector database — structured, traversable, temporally-versioned regulatory intelligence.

2.7M Nodes · 3 Pillars · Neo4j

Multi-Model Verification

No model validates its own output. Claude, Gemini, GPT-5, and Grok work in parallel. HMAC-signed sources prevent fabrication.

4+ LLMs · Cross-validation · HMAC

Research → Build → Run

Full vertical pipeline. Understand regulation, build compliant models with AI assistance, deploy as operational processes.

4 Workspaces · DAG Models

DPM-Native Reporting

97,481 reporting cells mapped across 695 templates in 21 frameworks. COREP, FINREP, IRRBB, ESG Pillar 3, and more.

21 Frameworks · 97K Cells

Validation architecture

Four minds. Zero self-validation.

No model checks its own work. BARK uses cross-provider diversity to eliminate single-model bias.

"What are the own funds requirements for institutions with significant trading book positions?"
CLAUDE
Claude Opus
Reasoning & orchestration
GEMINI
Gemini Pro
Research execution
GPT-5
GPT-5
Strategy diversity
GROK
Grok-4
Adversarial review
↓
6-Evaluator Validation Panel
GPT-5 H
Sonnet L
Grok H
GPT-5 L
Sonnet H
Grok L
✓ ALL FINDINGS VERIFIED — HMAC INTEGRITY CONFIRMED

Every assertion cryptographically linked to retrieved regulatory text.

See the difference

Same question. Different answers.

GENERIC AI
"What risk weight applies to residential mortgage exposures under CRR?"
Under CRR, residential mortgage exposures are assigned a risk weight of 35% as specified in Article 94(3). The capital conservation buffer of 3.5% also applies per CRD requirements.
Art. 94(3) — does not existBuffer is 2.5%, not 3.5%
~35%
BARK
"What risk weight applies to residential mortgage exposures under CRR?"
Under CRR Article 125(1), exposures fully secured by mortgages on residential property are assigned a risk weight of 35%, subject to conditions in Art. 125(2). Note: CRR II (2019/876) amends this with an exposure-to-value dependent approach.
CRR Art. 125(1)CRR II Art. 1(42)EBA/GL/2020/06
90%+

Process

Three steps to regulatory clarity

1

Ask a question

Describe your regulatory challenge in plain language. BARK decomposes it into a research plan with parallel investigation streams.

2

AI researches deeply

Multiple LLMs traverse the knowledge graph, cross-reference articles, and validate findings. Every claim traced to its source.

3

Build and deploy

Turn research into executable models, automated reports, and monitoring processes. Understanding to production in one platform.

0
Regulatory knowledge nodes
0%+
Accuracy on regulatory questions
0+
EBA & SFSA documents indexed
0
Reporting frameworks covered

Live system

Always on. Always ready.

BARK is not a tool. It's an operating system.

Knowledge Graph
2,700,000 nodes · 4 vector indexes · Neo4j
Document Store
1,249 documents · 195,000 chunks · ChromaDB
Research Engine
4 frontier models · 6 evaluators · HMAC
Build Workspace
DAG execution · Sheets / Python / Jupyter
Reporting Layer
21 frameworks · 695 templates · 97,481 cells

Built for banks

Security is the foundation.

We hold ourselves to the same standards we help banks meet.

EU Data Residency

All infrastructure in europe-north1 (Finland). Your data never leaves the EU.

Multi-Tenant Isolation

Isolated credentials per bank. Project-level workspace isolation.

Complete Audit Trail

Event-sourced workspace with hash-chained integrity. Every step tamper-evident.

HMAC Source Verification

Every finding cryptographically linked to retrieved text. Sources cannot be fabricated.

GDPREU data protection
EU Hostedeurope-north1
EncryptedTransit & at rest
IsolatedPer-bank credentials

Your regulation is complex.
Your tools shouldn't be.

Join the banks using BARK to turn regulatory complexity into competitive advantage.

Request a DemoExplore Product
✓ No commitment✓ 30-min demo✓ EU data only