> ## Documentation Index
> Fetch the complete documentation index at: https://docs.liberate.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Key concepts

# Key Concepts

## Core terminology and concepts for working with Liberate

Understand the foundational concepts before diving into Liberate's features.

## AI Agents

> [card] Voice AI
>
> AI agents that answer phone calls with human-like conversation. Handles inbound calls 24/7 with zero wait time—customers often can't tell they're speaking with AI.

> [card] Email AI
>
> AI agents that read, understand, and respond to customer emails. Automatically routes, replies, or escalates based on content and intent.

> [card] SMS AI
>
> AI agents that handle text message conversations. Supports two-way messaging for claims updates, policy questions, and service requests.

> [card] Digital Intake
>
> Web-based forms and chat interfaces powered by AI. Guides customers through FNOL submissions, quote requests, and policy changes.

***

## Customer Interactions

### Warm Transfer

When the AI agent hands off a call to a human agent, it's called a **warm transfer**. Unlike cold transfers, the human agent receives:

* Full conversation context
* Customer information pulled from your systems
* AI-generated summary of the request

This means customers never have to repeat themselves.

### Resolution vs. Escalation

| Term | Definition |
| - | - |
| **Resolution** | AI agent fully handles the request without human intervention |
| **Escalation** | AI agent transfers to a human agent for complex or sensitive requests |

Liberate tracks your **resolution rate**—the percentage of interactions handled entirely by AI. Most customers achieve 70-80% resolution rates.

### Call Flow

A **call flow** defines how the AI agent navigates a conversation:

```
Greeting → Identify Intent → Collect Information → Take Action → Confirm & Close
```

Call flows can branch based on customer responses, policy types, or business rules.

***

## Insurance Operations

### FNOL (First Notice of Loss)

The first report of an insurance claim. Liberate's AI agents can:

* Answer FNOL calls 24/7
* Collect all required claim information
* Create the claim in your claims system
* Provide the customer with a claim number

### Straight-Through Processing

When a claim or transaction is processed **entirely without human intervention**—from intake to completion. Liberate enables straight-through processing for routine claims by:

1. Collecting data via Voice/Digital AI
2. Validating against policy information
3. Creating records in your core system
4. Triggering downstream workflows

### Policy Servicing

Handling customer requests related to existing policies:

* Coverage questions
* Address/contact updates
* Payment inquiries
* Document requests
* Certificate of insurance

***

## Platform Architecture

### Environments

Liberate provides three environments for building and deploying:

| Environment | Purpose |
| - | - |
| **Preview** | Development and testing while you build |
| **QA** | Stakeholder validation before go-live |
| **Production** | Live traffic with real customers |

> [!WARNING]
>
> Always test in Preview first, validate in QA, then deploy to Production.

### Projects

**Projects** are containers that group related configurations together. Use projects to organize:

* Workflows for a specific line of business
* Configurations for different departments
* Separate client setups (for agencies/TPAs)

### Tenants

A **tenant** is your organization's isolated instance of Liberate. Your tenant ID appears in your platform URL (e.g., `yourcompany-ui.liberateinc.com`) and is automatically included in all workflow executions.

***

## Workflow Concepts

### Workflows

A **workflow** is an automated sequence of steps that processes data and connects systems. Workflows are triggered by events (incoming calls, form submissions, API requests) and execute without manual intervention.

```
Trigger → Process Data → Connect Systems → Complete
```

### Nodes

**Nodes** are the building blocks of workflows. Each node performs a specific action:

| Node Type | Purpose |
| - | - |
| **Start Event** | Entry point—receives incoming data |
| **Mapping** | Transforms data (combine fields, format values, restructure) |
| **Task** | Calls external systems (your AMS, claims platform, CRM) |
| **Decision** | Branches based on conditions |
| **End Event** | Marks workflow completion |

### Context

**Context** is the data object that flows through your workflow. Think of it as the workflow's memory:

```json theme={null}
{
  "tenantId": "acme-insurance",
  "environment": "production",
  "policyNumber": "HO-12345678",
  "claimant": {
    "firstName": "John",
    "lastName": "Smith"
  },
  "claimId": "CLM-98765"  // Added by Task node
}
```

* Every node can read from context
* Nodes add or modify data in context
* Context travels from node to node until workflow completes

### Connections

**Connections** are pre-configured integrations with external systems. Set up once, use across multiple workflows:

* Agency Management Systems (Vertafore, Applied, EZLynx)
* Core Systems (Guidewire, Duck Creek, Insuresoft)
* CRMs (Salesforce)
* Custom REST APIs

***

## Data & Security

### PII Masking

**Personally Identifiable Information (PII)** is automatically masked in logs and displays. You'll see `***` instead of sensitive data like names, policy numbers, and addresses. The actual data is processed securely—masking is for display only.

### Audit Trail

Every workflow execution, API call, and data access is logged. The **audit trail** provides:

* Who triggered the action
* What data was accessed or modified
* When the action occurred
* Execution status and errors

### Correlation ID

A **correlation ID** links related workflow executions together. When one workflow triggers another, they share the same correlation ID—useful for tracing complex processes across multiple workflows.

***

## Metrics & Analytics

### Key Performance Indicators

| Metric | Definition |
| - | - |
| **Resolution Rate** | % of interactions handled without human intervention |
| **Average Handle Time** | Time from call start to resolution |
| **Escalation Rate** | % of interactions transferred to human agents |
| **First Call Resolution** | % of issues resolved on the first contact |
| **Customer Satisfaction** | Post-interaction survey scores |

### ROI Tracking

Liberate calculates return on investment by measuring:

* Calls handled by AI vs. human agents
* Cost per interaction
* Time saved on routine tasks
* After-hours coverage value

***

## Next Steps

> [card] [Platform Overview](https://docs.liberate.ai/platform-overview.md)
>
> See how all the pieces fit together

> [card] [Voice AI Setup](https://docs.liberate.ai/voice-ai/setup.md)
>
> Configure your first AI agent

> [card] [Workflow Quickstart](https://docs.liberate.ai/workflows/quickstart.md)
>
> Build your first automated workflow

> [card] [Integrations](https://docs.liberate.ai/integrations/overview.md)
>
> Connect your existing systems


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.