​
Platform Architecture

Velatir consists of four main components that work together to provide comprehensive AI governance:

Policies

Define what needs review - GDPR, EU AI Act, bias detection, and custom rules

Flows

Control how decisions are routed based on conditions and risk levels

Review Tasks

The lifecycle of approval requests from creation to final decision

Channels

Where approvals happen - Slack, Teams, email, or web dashboard

​
How Components Work Together

AI Request Submitted

Your AI system makes a request through API, SDK, or integration

Policy Assessment

Advanced AI evaluates the request against GDPR, EU AI Act, bias detection, and custom policies

Risk-Based Routing

Low Risk → Auto-approved instantly
Medium/High Risk → Creates review task and routes via configured flow

Channel Delivery

Request sent to appropriate approval channel (Slack, Teams, email, or web)

Human Decision

Team member approves, declines, or requests changes with full context

Audit Trail

Every decision logged with reasoning, timestamps, and compliance status

​
Key Principles

​
1. Governance as an Enabler

Instead of blocking AI adoption, Velatir enables it by providing the oversight and compliance frameworks businesses need to deploy AI confidently.

​
2. Smart Automation

Only human review what actually needs human input. Our AI-powered policy engine automatically handles routine decisions while escalating complex or risky ones.

​
3. Work Where You Work

Approvals happen in your existing tools - no need to constantly check another dashboard.

​
4. Full Transparency

Every decision includes the reasoning, risk assessment, and compliance status. Perfect for audits and continuous improvement.

​
5. EU-First Compliance

Built with GDPR and EU AI Act compliance as core requirements, not afterthoughts.

​
Decision Flow

Every AI request follows this pattern:
  1. Request Created - AI system makes a request through API, SDK, or integration
  2. Policy Evaluation - Advanced AI assesses the request against all active policies
  3. Risk Scoring - Request gets a risk level (Low, Medium, High, Critical) with confidence scores
  4. Flow Matching - System finds the appropriate flow based on conditions
  5. Routing Decision - Auto-approve or route to appropriate approval channel
  6. Human Review (if needed) - Approve, decline, or request changes
  7. Action & Logging - Decision executed and logged for audit trail

​
Request Lifecycle States

1

Pending

Request just created, not processed yet
2

Processing

Being evaluated against policies and flows
3

Approved

Automatically approved based on policy assessment
4

Requires Intervention

Waiting for human approval in designated channel
5

Rejected

Human reviewer declined the request
6

Change Requested

Human reviewer wants modifications before approval

​
Integration Options

Velatir integrates into your existing AI workflows through:
  • Direct API - REST endpoints for custom integrations
  • SDKs - TypeScript/JavaScript and Python libraries
  • n8n Node - Official node for workflow automation
  • MCP Server - Model Context Protocol for AI agents
  • Webhooks - Event-driven notifications and responses

Explore Policies

Learn how AI requests are evaluated

Understand Flows

See how decisions are routed

Review Tasks Deep Dive

Follow the approval process