Multi-agent orchestration
Several virtual employees work together: the customer contact agent passes a return to the order processor, which updates accounts payable.
Antonic.ai platform
Sovereignty you can audit
Antonic.ai is the multi-agent platform behind every virtual employee. Model-agnostic, connected to 400+ systems and designed to prove where your data is and what the agent did.
The sovereignty ladder
Five levels. The right choice per use case.
A public FAQ needs something different from an agent handling citizen or client files. Antonic.ai delivers up to level 4, and the logs show which level you get.
- Level 4Dutch infrastructure
Open models on Dutch hardware, fine-tuning without data leaving the country.
Antonic.ai - Level 3European hosting, open or European model
Mistral, Llama or GPT-NL on European infrastructure. No data to the US.
Antonic.ai - Level 2European vendor, US model
The platform is in Europe, but prompts and data go to a US language model.
About 95% of Dutch AI vendors - Level 1US cloud, EU region
Data in the EU, but a US parent: legal access and switch-off risk remain.
Hyperscaler EU regions - Level 0US SaaS
Data and models in the US, fully under the CLOUD Act.
Default for many large platforms
Under the hood
What every virtual employee gets as standard
Model-agnostic
The right model per use case: open, European or — where allowed — a commercial model. Switch without rebuilding.
400+ connectors
n8n-native: AFAS, Exact, SAP, Salesforce, Pipedrive, TOPdesk, Microsoft 365, WhatsApp and your own APIs.
Audit trail per action
Every step recorded: what the agent saw, decided and did, with which model and where.
Human in the loop
Limits per role, approval steps and instant handover to a colleague with a summary.
Exit plan as standard
With every delivery: how you switch to another model or vendor. Aligned with Dutch government cloud policy.
AI Act transparency
Agents disclose that they are AI by default (Art. 50) and mark generated content where required.
Regulation
Ready for what applies now — and what is coming
- AI Act Art. 50: transparency for chatbots and agentsin force
- Dutch Cybersecurity Act (NIS2) for ~8,000 organisationsin force
- Government cloud policy: EEA storage, encryption, exit planin force
- Marking of synthetic content for existing systems
- High-risk AI: recruitment, credit scoring, education
Integrations
Works with the systems you already have
- AFAS
- Exact
- Twinfield
- SAP
- Microsoft Dynamics
- Salesforce
- Pipedrive
- HubSpot
- TOPdesk
- ServiceNow
- Microsoft 365
- Google Workspace
- Zendesk
- Shopify
- Magento
- Case systems
- Your own APIs
Want to know which sovereignty level your processes need?
In 30 minutes we map which process fits best, what it delivers and how your data stays sovereign. No sales pitch — an honest assessment.
- Confirmed within one working day
- Video or on-site
- First ROI estimate on the spot