EU AI Act: high-risk obligations on December 2, 2027
Sovereign AI Governance for governments and large organisations
Your data never leaves your infrastructure. The 100% on-premise platform to comply with the EU AI Act, GDPR and ENS: inventory, assessment, monitoring and compliance, with an AI copilot that operates everything by natural language, in one place.
European company · European infrastructure · Full data sovereignty · On-premise
EU AI ActGDPR / RGPDENSNIS2ISO 42001
su-dominio.gob.es/dashboard
Executive dashboard
Last 14 days · Updated 5 min ago
Export14 días ▾
AI Systems+3 this month
24
High Risk-1 vs previous
3
Compliance+4pp
76%
LLM Cost-12% savings
€847
Recent activity
PeticionesCumplimiento
Key metrics
Compliance gap76%
Fairness score0.87
Policies OK6/8
XAI Reports9
Open incidents2
Recent activitySee all →
GPT-4 Legal Assistant classified as High Risk2h ago
Fairness evaluation completed for Sentencing Model5h ago
DRIFT_CHECK policy triggered on Prediction Model8h ago
Key dates
The calendar is no longer a single date
The Digital Omnibus postponed the high-risk rules and left transparency untouched. The result is a staggered calendar in which what binds you today is not what almost everyone has written down.
Kairos projects these dates onto your inventory and tells you which specific systems fall into each stage.
59
Integrated modules
13
Regulations covered
100+
Mapped requirements
100%
On-premise
Urgency
Why act now
Regulation, operational risk and sovereignty are now part of the same conversation.
Already in force
The transparency obligations of Article 50 have been binding since 2 August 2026 and the Digital Omnibus did not postpone them: disclosing that people are interacting with an AI, marking synthetic content and declaring published texts that inform on matters of public interest. On 2 December 2026 two new prohibited practices come into force as well.
December 2027
The Digital Omnibus postponed the high-risk obligations (Annex III) from August 2026 to December 2, 2027, and product-safety AI to August 2028. More room to do it right, not to wait: identifying and registering high-risk systems remains an obligation, and a human reviewer does not automatically take a system out of Annex III scope.
Up to 35M EUR
Fines for non-compliance with the EU AI Act, or 7% of annual global turnover. For the public sector, sanctions include asset liability and disciplinary proceedings.
13 regulations
EU AI Act, GDPR, ENS, NIS2, ISO 42001, ISO 27001, LOPDGDD, NIST AI RMF, AI Liability Directive, DORA, EN 18286:2026, ISO/IEC 42005 and the Spanish Organic Law on AI governance (2026). Kairos covers them all with 100+ mapped requirements.
Sectors
Built for organisations
that cannot afford to fail
Kairos deploys where your data lives. No external clouds, no sovereignty risks, no third-party dependencies for your most sensitive information.
Public Administration
City councils, ministries, regulatory bodies and European agencies
Banking & Insurance
Credit scoring, fraud detection, risk models and AI-powered customer service
Healthcare
Assisted diagnosis, triage, clinical records management and predictive models
Energy & Infrastructure
Predictive maintenance, grid optimisation and demand models
Large Enterprises
Any organisation with more than 10 AI systems in production
Consulting & Audit Firms
Firms advising their clients on AI regulatory compliance
Pharmaceuticals & Life Sciences
Model validation under GxP, pharmacovigilance, drug discovery and clinical trials, with traceability for EMA, FDA and GAMP 5
Platform
Six pillars of governance
From registry to explainability. Each pillar covers a critical dimension of AI governance required by the EU AI Act.
Operate the entire platform by talking, with human control
An integrated AI copilot (Cmd+K) that understands Kairos: queries status, recommends what is missing and executes actions by natural language. Every write goes through a human confirmation gate, and the copilot itself is governed by the platform (inventoried, with guardrails and audited).
Query, recommendation and execution by natural language in 5 languages
Human confirmation on every action that modifies data
Operates end to end: inventory, risks, use cases, approvals, lifecycle and controls
Answers regulatory queries (EU AI Act, GDPR, ISO 42001) grounded in the regulatory knowledge base
The copilot governs itself: registered in the inventory, with guardrails and traced in the audit trail
Swappable AI engine, on-premise or sovereign
99
Actions
5
Languages
100%
Human gate
59
Modules
Copiloto KairosCmd+K
¿Qué le falta al caso de uso "Scoring Crediticio" para cumplir?
Revisó la completitud de gobernanza
Le faltan 2 cosas: la FRIA y el nivel de riesgo. ¿Abro la FRIA y propongo riesgo ALTO?
Confirmar acción
Crear FRIA para "Scoring Crediticio"
Aprobar
Rechazar
Experience
Governance that guides you
Not an endless form: a copilot. The platform tells you what is missing, takes you to fix it and shows you how everything connects.
Living governance map
The whole network of use cases, systems, risks and assessments in one interactive graph. What needs attention is visible from afar.
5-dimension health signal
Every entity shows its completeness: definition, risk, compliance, validation and approval. One glance, zero doubts.
3
4
5
IAAlto riesgo · Anexo IIIConfianza alta
AplicarDescartar
AI-guided onboarding
Describe your use case and the AI suggests classification, risks and obligations. The path adapts to the EU AI Act risk level.
Revisión inicial
2 votos · 12 mar
Comité de IA
3 votos · 28 mar
Despliegue a producción
2/4 votos · en curso
Traceable approvals
Visual workflows per risk level: who approved each step, on what date, and what is pending. Audit-ready.
Copiloto KairosCmd+K
¿Qué le falta a este caso de uso para cumplir?
Faltan la FRIA y el nivel de riesgo. ¿La creo?
Confirmar acción
AprobarRechazar
Copilot that acts for you
Ask it in natural language (Cmd+K) to query, recommend or execute. Every change goes through your confirmation, and the copilot itself is governed by the platform.
Capabilities
59 modules. One platform.
Every module is functional, interactive and connected to the rest of the platform.
Use cases
How Kairos is used in practice
10 real-world scenarios with professional profiles from the public sector and large organisations. From onboarding to AESIA inspection.
Compliance
Multi-regulation from day one
Complete requirement mapping, impact assessments, EU AI Act conformity assessment and Art. 73 notifications with pre-configured authority directory.
EU AI Act
Core
EU Regulation
2024
GDPR / RGPD
Core
EU Regulation
2016
ENS
Royal Decree
2022
NIS2
EU Directive
2022
ISO 42001
Core
Standard
2023
ISO 27001
Standard
2022
NIST AI RMF
Framework
2023
DORA
Core
EU Regulation
2022
LOPDGDD
Organic Law
2018
AI Liability
EU Directive
2022
Ley Gobernanza IA (ES)
Core
Spanish Organic Law
2026
Art. 73 — Incident notification
Automatic draft with legal deadlines and authority directory: AESIA, AEPD, CNMC, CCN-CERT, INCIBE and the European AI Office.
Positioning
Kairos vs. other solutions
The best-known AI governance platforms are SaaS, cloud-only and with entry costs exceeding 50,000 EUR/year. Kairos offers a real alternative.
Positioning
Kairos
Others
Real on-premise deploymentYour data never leaves your infrastructure. No third-party dependencies for storage.
Included
Cloud only (SaaS)
Integrated LLM GatewayOpenAI-compatible proxy with PII detection, data residency policy, rate limiting, budgets and automatic fallback.
Included
Not included
Entry costAccessible licence for public administrations and mid-sized companies, with no per-user cost.
Included
From 50,000 EUR/year
LLM-as-Judge evaluationsAutomated tests for bias, hallucinations, PII and prompt injection using AI as a judge.
Included
Not available
Pre-configured Art. 73Directory of Spanish and European authorities, automatic drafts and deadline tracking.
Included
Manual setup
Automated Policy-as-CodeSix types of executable policy running every 5 minutes. Evaluates thresholds, drift, conformity and approvals.
Included
Partial
AI BoM + Supply Chain RiskBill of Materials with 9 component types and supplier risk assessment across 5 dimensions with sovereignty alerts.
Included
Partial
European company, EU jurisdictionKairos is a European company. Its competitors (Credo AI, Holistic AI, IBM) are US companies subject to the CLOUD Act and FISA.
Included
US company (CLOUD Act)
Sovereignty
European digital sovereignty for real
A data centre in Europe is not sovereignty if the company operating it is American. Kairos is a European company, with European jurisdiction, with no legal backdoors.
The problem Europe ignores
And it affects every public administration and large enterprise
CLOUD Act (2018)
Requires any US company to hand over data stored anywhere in the world if requested by a US court or agency — even if the data is on a server in Frankfurt or Madrid.
FISA Section 702
Allows the NSA to access data of non-US citizens without a warrant. Applies to Microsoft, Google, Amazon, Oracle and any cloud provider subject to US jurisdiction.
Schrems II (CJEU, 2020)
The Court of Justice of the EU invalidated the Privacy Shield. It ruled that data transfers to the US do not offer guarantees equivalent to the GDPR. The issue remains unresolved.
“Servers in Europe” is not enough
AWS Frankfurt, Azure Madrid or Google Belgium are still US companies. They are legally required to comply with the CLOUD Act, regardless of where their servers are physically located.
The Kairos answer
European company. European jurisdiction. No exceptions.
European company
Kairos is a company incorporated in the European Union, subject exclusively to European law. No foreign government can demand that we hand over your data.
Sovereign infrastructure
Deployed in your own data centre or on European infrastructure. No dependence on US hyperscalers. Your data never leaves EU jurisdiction.
Native compliance
Designed from day one for the EU AI Act, GDPR, ENS and NIS2. Not a later adaptation of an American product for the European market.
US cloud vs. Kairos
Applicable jurisdiction
US law (CLOUD Act, FISA, PATRIOT Act)
EU law exclusively
Government data access
US can demand data without notifying the owner
Only with European court order and notification
International transfer
Data may be transferred to the US by legal request
Data never leaves your infrastructure
Real GDPR compliance
Incompatible per CJEU (Schrems II)
Native — designed for GDPR from the start
Code audit
Closed and proprietary code
Full access to code deployed on your infrastructure
Technological independence
Vendor lock-in with foreign hyperscaler
No external dependencies, portable, sovereign
“Digital sovereignty is not a luxury, it is a strategic necessity. Europe cannot govern its artificial intelligence with tools that are subject to a foreign country's jurisdiction.”
— Kairos founding principle
Architecture
Built for the public sector and large enterprises
On-premise, multi-tenant and secure. Compatible with ENS, GDPR and public procurement regulations.
Fully on-premise
Deployed in your data centre. No external cloud dependencies. Local PostgreSQL. Meets ENS data sovereignty requirements.
Secure multi-tenant
5 roles (Admin, Governance Manager, Model Owner, Auditor, Viewer) with JWT and cross-org superadmin. RLS on all tables.
Full observability
Immutable audit trail protected by PostgreSQL triggers. Real-time PII detection. Alerts and automated cron.
Security by design
Immutable audit trail protected by PostgreSQL triggers
SHA-256 hash for all Gateway API keys
Real-time PII detection: email, national ID, IBAN, credit cards
Jailbreak and harmful content detection in prompts
RBAC with 5 roles and organisation isolation via JWT
Webhooks signed with HMAC-SHA256 and automatic retries
Full data sovereignty — 100% on-premise deployment
Don't wait for the inspection to start governing your AI
Request a personalised demo and discover how Kairos can prepare your organisation for the EU AI Act in weeks, not months.