Security and data
Everything you’d ask before
trusting us with your operations
Learn where your data lives, who can touch it, and what running OAIZ looks like in month two. Straight answers.
Trust and control
Set the steps, permissions and data rules for your automations. Review the evidence behind the platform.
Reliable execution
Build workflows with explicit steps, conditions and approvals. Run real inputs in a draft and inspect each step before publishing.
Test in a draft. Keep a version history. Review the result before going live.
Explore workflow controls
Example workflow
- Incoming email
- Draft a reply
- Approval
Control what reaches AI
Keep data retrieval, AI processing and actions separate. Pass each model only the fields its task needs.
Choose the inputs for each AI task. Keep the rest of the process in the workflow.
See how data moves through a workflowExample field selection

- Order ID
- Passed to AI
- Request
- Passed to AI
- Email address
- Not passed
Control who has access
People work within their permissions. Agents have their own identity, with grants for the integrations and actions they need.
Set read and write access. Put an approval before a sensitive step.
Explore access controlsExample access policy

| Resource | Read | Write |
|---|---|---|
| Helpdesk | Allowed | Allowed |
| CRM | Allowed | Blocked |
| Billing | Blocked | Blocked |
Protect sensitive data
Detect sensitive text and replace it with tokens. Apply rules to how that information can be used.
Decide what reaches the model and where sensitive information can be used in the workflow.
Explore sensitive-data controlsExample text replacement

Original text
Alex Morgan · alex@example.com
Text sent to the model
[PERSON_1] · [EMAIL_1]
Choose where it runs
Match model providers and deployment options to your data requirements. Check where each part of the work runs.
Primary application data is stored in Belgium. AI providers and shared services can process data outside the EU.
Review processing locations
- Primary app data
- Belgium
- AI processing
- Depends on provider and model
- Shared services
- Can process data outside the EU
Security you can verify
Review the security controls, assessment results and supporting documents behind OAIZ. Check the status of each before you decide.
Code, dependencies and secrets are checked on every change. Connections and stored data use encryption.
Review security documentationSecurity evidence

- ISO 27001
- In progress
- CASA Tier 2
- 9.7 / 10
- Penetration test
- September 2026
- GDPR
- Data processing agreement
Swedish company.
European primary infrastructure.
We are based in Sweden, Nordic funded and owned. Primary application data is stored in Belgium. AI providers and shared services can process data outside the EU. See processing locations for an overview and DPA Schedule 3 for the governing disclosures.
You own the data. We look after it.
You connect your systems to OAIZ and stay the data controller. We act as your sub-processor under a data processing agreement, and handle encryption, backups, failover and the engineers on call.
- Company
- Oaiz AB, Sweden · Nordic funded and owned
- Hosting
- Primary application: Google Cloud europe-west1, Belgium
- Your data at rest
- Primary application storage: Belgium, encrypted
- Language models
- Locations depend on the provider and selected model. Processing locations and available models.
- Our role
- Sub-processor under your DPA
- Backups
- Daily
- Breach notice
- Within 48 hours, per our DPA
What month two looks like
The integration is live. Here is what keeps running on its own, what you can change safely, and what is still on your team’s plate.
It keeps working when you close the lid
Nothing runs on your laptop. Events from your systems trigger workflows in our cloud - at 05:14 on a Sunday as reliably as at 10:00 on a Tuesday.
Connect
One line of code to send us an event.
Use a native integration, or push events from anything you run. Whoever owns the data decides what flows in.
oaiz.events.send("order.created", order)Change and Prove
Change it safely. Show how it decided.
Full Versioning
Branch, test and compare before you publish. Revert in one click.
Every step on the record
What the agent saw, decided and did - the trail you need for AI Act record-keeping.
Code when agents aren’t enough
Drop into a sandboxed code step for the hard parts. Less AI in your AI automation.
Run and Support
Know what it costs. Know who’s watching.
Budgets you set, spend you can see
Limits per workflow and cost per run. Work stops before it overshoots.
Monitored around the clock
We watch the platform 24/7, and you can set up notifications for failing workflows.
People, not a ticket queue
A named customer success contact. Premium SLA with 24/7 support when you need it.
Your workload in month two
The short list is yours. The long list is ours, whether you build alone or with us.
- 1.What to automate next
- 2.Who gets access, and approvals
- 3.Custom integrations - or hand them to us
Infrastructure, scaling and uptime
Backups and failover
Keeping native integrations working
Security patches and dependencies
New model versions
24/7 platform monitoring
Delivery & Implementation for Enterprise
Workflow SLAs for Enterprise
Build with us or by yourself
Book a call if you want help from us, or try for free to get started on your own