Skip to content

Technology ​

BlueUP Platform: Layered Architecture

BlueUP is not a standalone product: it's an integrated platform in three layers designed so regulated organizations can adopt agentic AI without compromising security, compliance, or sovereignty.

Platform architecture ​

LAYER 3: SOVEREIGN EXECUTION
BlueUP Core (Financial engine)
150,657 journals/sec
Native multi-GAAP
MCP (Model Context Protocol) is an open standard, hosted by the Agentic AI Foundation at the Linux Foundation, that standardizes how AI applications access external tools and data. In banking and insurance it is where the host controls what an agent invokes. At BlueUP, MCP gateway governance is on the roadmap.Read more → MCP / LLM Gateways (in design)
Tool and model
governance
LAYER 2: COMPLIANCE AND GOVERNANCE
ComplianceView (Compliance monitoring)
Continuous tracking
98 controls
BlueUPALM (Anti-money laundering (AML) is the prevention of money laundering and terrorist financing. In Spain, Law 10/2010 requires credit and financial institutions to apply customer due diligence and to report suspicious transactions to SEPBLAC; Regulation (EU) 2024/1624 harmonizes it from July 2027.Read more → AML/DORA (Digital Operational Resilience Act) is Regulation (EU) 2022/2554 on digital operational resilience for the financial sector. It requires EU banks, insurers and investment firms to withstand ICT disruptions and threats, respond to them and recover. It has applied since 17 January 2025.Read more → DORA)
Banking-grade regulatory
compliance
LAYER 1: ZERO TRUST REACHABILITY
BlueUP Connect (Zero Trust desktop client)
Identity-first access
to authorized services
OpenZiti is NetFoundry's open-source connectivity substrate: X.509 identity, end-to-end encryption and dark services with no inbound ports or public IP. VPN replacement is one of its use cases: identity is authorized before any data path exists. At BlueUP it underpins Zero Trust reachability.Read more → OpenZiti / NetFoundry is the company that created and maintains OpenZiti, the open-source Zero Trust connectivity substrate, and counts Cisco Investments among its investors. A regulated entity uses it to keep services off the public internet. BlueUP is an official partner offering self-hosted deployment or managed connectivity.Read more → NetFoundry
Identity-first connectivity
Dark services by default
SUBSTRATE: CRYPTOGRAPHIC IDENTITY
Keycloak (humans) · SPIRE (the SPIFFE Runtime Environment) is a production-ready implementation of the SPIFFE APIs and a CNCF graduated project: it issues each workload an SVID, a verifiable identity in X.509 or JWT form. In banking and insurance it replaces static credentials with short-lived, auto-rotating workload identity.Read more → SPIRE SVIDs (workloads) · Biscuit is an open-source authorization token from the Eclipse Foundation, verified with public keys and attenuable offline: a narrower token derives from another with no call to the issuer. In banking and insurance it bounds delegation between services. BlueUPALM issues and attenuates it; BlueUP Core has it in design.Read more → Biscuit Tokens (offline authorization) · OPA (Open Policy Agent) is an open source policy engine and a graduated CNCF project that decouples policy decisions from their enforcement and expresses them in the Rego language. In banking and insurance it allows access rules to be versioned and audited as code.Read more → OPA (infra) · Cerbos (app)

Layer 1: Zero Trust reachability ​

The fundamental principle: without a valid cryptographic identity, no data path exists.

Services published over OpenZiti have no public IP, don't respond to port scans, don't appear on Shodan. They only "exist" for authenticated identities with matching policy.

OpenZiti

Open-source connectivity substrate: encrypted tunnels, dark services, service policy.

BlueUP Connect

Desktop client showing users only their authorized services.

Cryptographic identity

Every agent, service and human has a verifiable cryptographic identity.

Dark services

No inbound ports, no public IP, invisible to the internet.

Technology Partner: NetFoundry

Our connectivity substrate is built on OpenZiti, the open-source platform developed by NetFoundry. As an official NetFoundry partner, we offer both self-hosted deployment and managed connectivity for customers who require it.

NetFoundry is backed by investors including Cisco Investments and partners like Stellar Cyber and Intrusion for high-security environments.


Layer 2: Compliance and governance ​

Continuous monitoring and compliance capabilities on the same platform, not bolted on.

BlueUPALM

Banking-grade AML/DORA compliance: screening, incident management.

ComplianceView

98 controls aligned with NIST (National Institute of Standards and Technology) is the US federal agency within the Department of Commerce that publishes cybersecurity standards such as the Cybersecurity Framework and the SP 800 family. Banking and insurance adopt them as a control catalog; ComplianceView maps its controls to NIST SP 800-53r5.Read more → NIST, ISO 27001 (ISO/IEC 27001:2022) is the international standard from ISO and IEC with the requirements for an information security management system (ISMS) and its reference controls. An accredited body audits and certifies conformity, the evidence banking and insurance use to demonstrate risk control.Read more → ISO 27001, DORA, and FINOS (Fintech Open Source Foundation) is the Linux Foundation umbrella organization that unites financial services to build open technology and standards. It hosts the Common Domain Model (CDM) and the SDLC Controls working group, whose controls ComplianceView integrates alongside NIST, ISO 27001 and DORA.Read more → FINOS. Automated collectors.

OPA

Centralized access and infrastructure policy evaluation.

Cerbos PDP

Contextual ABAC/RBAC authorization for business logic.


Layer 3: Sovereign execution ​

Business logic runs on controlled infrastructure with institutional-grade performance.

BlueUP Core (financial engine in Rust)

11 crates, 1,320 tests, multi-GAAP accounting (Sectoral/IFRS/Tax), 150,657 journals/sec (benchmark of 2026-07-02: 10,000 events with three journals each, in-memory, on a Mac mini with a 6-core Intel Core i5-8500B and 8 GB).

MCP gateway (in design)

Layer in design, not implemented today: it will govern which tools agents can invoke, by identity and policy.

LLM gateway (in design)

Layer in design, not implemented today: it will control access to language models with human approval points.

gVisor is an open-source container sandbox: an application kernel in user space intercepts system calls and separates the workload from the host kernel. In banking and insurance it limits the blast radius of a compromised container. BlueUP Core declares it as the runtimeClass in its Kubernetes manifest.Read more → gVisor sandbox

Kernel-level isolation of the application services: runtimeClass gvisor on BlueUPALM (the triage agent included), on ComplianceView and on OPA.


Identity substrate ​

The entire system shares a unified identity model:

LayerTechnologyFunction
HumansKeycloak OIDC/PKCEFederated authentication without static passwords
WorkloadsSPIRE SVIDs (X.509 or JWT)Rotating cryptographic identity per service
AuthorizationBiscuit TokensAuthorization tokens with offline attenuation
EncryptionmTLS on the OpenZiti overlay; WireGuard between cluster nodesMutual verification on every overlay connection

Complete technology stack ​

LayerTechnologies
FrontendReact, TypeScript, CSS Modules
BackendRust (Axum), Python (FastAPI), NATS JetStream
SecurityOpenZiti, Keycloak, SPIRE, OPA, Biscuit Tokens, Cerbos
AI & DataExternal model provider (triage), MCP SDK; local or regional provider (vLLM / Ollama) in design
InfrastructureGoogle Cloud, Kubernetes (Talos Linux), Terraform, Gitea Actions
IsolationCilium (network), gVisor (kernel), eBPF (observability)

Design principles ​

Identity as perimeter

Security doesn't depend on server location, but on verifiable cryptographic identity.

Dynamic privileges

AI proposes, but OPA policies and human intervention act as security brakes.

Minimum blast radius

A compromise in one agent never translates to a systemic breach.

Compliance as an architectural requirement

DORA, AML, The AI Act is Regulation (EU) 2024/1689, the European risk-based framework for AI: it bans unacceptable uses, regulates high-risk systems and requires transparency from general-purpose models. Its high-risk regime covers credit scoring and life and health insurance pricing from 2 December 2027 (Reg. 2026/1744).Read more → AI Act and The GDPR (General Data Protection Regulation) is Regulation (EU) 2016/679 on data protection, applicable since 25 May 2018. It reaches every entity established in the EU and anyone outside it offering goods or services to people in the Union or monitoring their behavior. Fines run up to 20,000,000 EUR or 4%.Read more → GDPR are architectural requirements, not add-ons.

AI provider

Triage uses an external model provider; a local or regional provider is in design.


Application by industry ​

This architecture is applied differently depending on the regulated industry it serves. See operational details in:

  • Private Banking — Enhanced KYC (Know Your Customer) is the sector name for customer due diligence under Directive (EU) 2015/849: identifying and verifying the customer, identifying the beneficial owner, assessing the relationship's purpose and monitoring it over time. It is completed before the relationship opens, with exceptions (Article 14).Read more → KYC, continuous screening and traceability for HNWI.
  • Insurance — Insurance-specific AML with integrated SEPBLAC (Servicio Ejecutivo de la Comisión de Prevención del Blanqueo de Capitales e Infracciones Monetarias) is Spain's Financial Intelligence Unit and its supervisor for the prevention of money laundering and terrorist financing. Obliged entities file suspicious transaction reports with it (Law 10/2010).Read more → SEPBLAC workflow.
  • Fintech & Agentic AI — Zero Trust is the security model NIST formalizes in SP 800-207: network location grants no implicit trust, and every access is authenticated and authorized separately. In banking and insurance it limits lateral movement after a credential is stolen; BlueUP applies it with per-service cryptographic identity.Read more → Zero Trust governance for autonomous agents.

Reference material: DORA 2026 guide and DORA calculator.


Want to dive deeper into the architecture?

Download our technical whitepaper on identity-first architecture for agentic AI.

→ Download whitepaper | Request demo

Last updated: