Izere/Platform

The execution
intelligence layer.

Izere ingests programme data from existing institutional systems, models actors and dependencies as a dynamic graph, and produces explainable risk intelligence with a continuous audit trail. Five layers. API-first. Federated by default.

System architecture

Five layers, one evidence chain.

Each layer is independently deployable and independently auditable. Data flows are versioned end-to-end so any score can be reproduced from raw inputs at any point in time.
izere/architecture · v2.4.0 SCALE 1:1 · UTC 2026-04-30
L1 · SOURCES TED · PPDS Procurement notices Arachne Risk scoring Cohesion OPs Programme reports RRF Scoreboard Milestones · targets Member State APIs National systems L2 · INGEST Unified data model Schema · entities · events programmes contracts · actors milestones · events disbursements governance graph L3 · MODEL Graph neural net Actor · dependency · risk edges: 2.4M nodes: 487K trained: federated L4 · EXPLAIN SHAP · audit Per-score attribution +18.4 Bidder conc. +14.2 Coordination +11.8 Financing gap +9.1 Milestone amend. +7.3 Governance dist. Σ = 84/100 contestable · versioned L5 · SURFACES Dashboard PM · Auth. Alert feed Webhook · email Audit pack Reproducible PDF REST · GraphQL Programmatic Briefing report Monthly · PDF AUDIT LANE · L0 Versioned lineage · model cards · access logs · evidence chain Spans every layer · reproducible to raw input · EU AI Act high-risk system documentation generated continuously
L1 Sources Public-data feeds and institutional APIs. No proprietary integration required.
L2 Ingest Unified data model with versioned schema. Raw inputs preserved for reproducibility.
L3 Model Graph NN trained federated; insights compound without raw data leaving perimeter.
L4 Explain SHAP attribution per score. Every output is contestable by an institutional auditor.
Data sources

Read what programmes already produce.

Izere does not ask institutions to change their procurement, programme-management, or supervision systems. We integrate against existing public and institutional data feeds.
EU.001

TED · PPDS

EU procurement notices. Contract awards, modifications, contracting authority structure, supplier identity.

EU.002

Arachne

Commission risk-scoring database. Supplier-level fraud and irregularity signals, beneficial-ownership network.

EU.003

Cohesion OPs

Operational programme reports. Absorption rate, milestone status, financial execution, performance framework.

EU.004

RRF Scoreboard

Recovery and Resilience Facility milestones and targets. Disbursement requests, fulfilment ratings, reform conditioning.

MS.001

Member State APIs

National managing authority systems. Contract-level execution data, governance changes, sub-recipient reporting.

EU.005

ECA reports

European Court of Auditors special and annual reports. Findings, recommendations, observations, follow-up status.

PUB.001

Procurement registries

National-level procurement systems beyond TED scope. Below-threshold contracts, framework call-offs, regional authorities.

REG.001

Beneficial ownership

UBO registries across EU-27. Cross-border ownership chains, ultimate beneficial owners, sanctions cross-reference.

Integration surfaces

Three ways to consume Izere.

Choose the surface that fits your institutional process. All three sit on the same evidence chain — programme managers, auditors, and policy leadership see the same scores.
SURFACE 01

Dashboard UI

Programme managers and managing authorities. Real-time portfolio view, programme detail with full evidence chain, what-if intervention modelling, exportable audit packs.

https://app.izereai.com/portfolio
SURFACE 02

REST + GraphQL API

Programmatic access for institutions integrating Izere into existing supervision pipelines. Risk scores, signals, audit metadata. Webhook subscriptions for threshold events.

POST /v2/programmes/:id/score
SURFACE 03

Briefing reports

Monthly PDF briefings for policy leadership and audit committees. Portfolio summary, top-risk programmes, intervention recommendations, methodology notes.

monthly · institutional · branded
Specifications

Built for institutional deployment.

These are the constraints Izere ships against. Reviewable in detail by institutional security and procurement before contracting.
Hosting EU sovereign cloud · OVHcloud (FR), Open Telekom Cloud (DE), or institutional on-premise. No US-jurisdiction routing at any layer. EU-only
Data residency All raw and derived data resides within the institution's selected EU jurisdiction. Federated training emits gradients only — never raw data. Federated
Encryption AES-256 at rest, TLS 1.3 in transit, customer-managed KMS keys optional. Confidential compute for sensitive programmes. CC available
Compliance GDPR Art. 28 DPA, EU AI Act high-risk system documentation, ISO 27001 (in audit), SOC 2 Type II (Q3 2026 target). Auditable
Authentication SAML 2.0, OIDC, eIDAS-compliant institutional identity. SCIM provisioning. MFA enforced. Granular role-based access by programme. eIDAS
SLA 99.9% availability for dashboard and API. P95 score-query latency < 250ms. Recovery time objective 4h, recovery point objective 15min. 99.9%
Audit trail Every score reproducible end-to-end from raw inputs. Append-only event log. Independent third-party auditor read-access available. Reproducible
Procurement Compatible with EU framework agreement procedures. DG REGIO, DG GROW, ECA, EIB pre-qualified vendor list (in progress). Pre-qual.

Need a technical walkthrough?

We provide architecture deep-dives for institutional security, audit, and procurement teams. Reviewable documentation in advance. Two-hour technical session.

Request technical session →