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

# Expense Auditor

> Build an AI agent that reviews expenses, identifies anomalies, ensures compliance, and generates audit reports

## Overview

An Expense Auditor powered by Pylar reviews expenses, identifies anomalies, ensures policy compliance, and generates audit reports to maintain financial integrity.

## What the Agent Needs to Accomplish

The agent must:

* Review expense submissions
* Identify anomalies and outliers
* Check policy compliance
* Generate audit reports
* Track approval workflows
* Identify fraud patterns

## How Pylar Helps

Pylar enables the agent by:

* **Unified Expense View**: Combining expenses, policies, and approvals
* **Anomaly Detection**: Automated anomaly identification
* **Compliance Checking**: Policy compliance validation
* **Audit Reporting**: Automated audit report generation

```mermaid theme={null}
graph LR
    subgraph "Data Sources"
        EXPENSES[Expenses]
        POLICIES[Policies]
        APPROVALS[Approvals]
    end
    
    PYLAR[Pylar<br/>Unified Views]
    
    TOOLS[MCP Tools<br/>detect_anomalies<br/>check_compliance<br/>generate_reports]
    
    AGENT[Expense Auditor]
    TEAM[Finance Team]
    
    EXPENSES --> PYLAR
    POLICIES --> PYLAR
    APPROVALS --> PYLAR
    PYLAR --> TOOLS
    TOOLS --> AGENT
    AGENT --> TEAM
    
    style PYLAR fill:#FF4017,stroke:#CC3300,color:#fff
```

## Without Pylar vs With Pylar

### Without Pylar

**Challenges**:

* ❌ Manual expense review
* ❌ Time-consuming anomaly detection
* ❌ Complex compliance checking
* ❌ Limited audit capabilities

**Implementation Complexity**: \~5-6 weeks

### With Pylar

**Benefits**:

* ✅ Automated expense review
* ✅ Real-time anomaly detection
* ✅ Policy compliance automation
* ✅ Efficient audit reporting

**Implementation Complexity**: \~6-7 hours

## Step-by-Step Implementation

### Step 1: Connect Data Sources

1. **Connect Expense System** (Expense reports, receipts)
2. **Connect HR System** (Employee data, policies)
3. **Connect Approval System** (Approvals, workflows)

### Step 2: Create Expense Views

**Expense Anomaly View**:

```sql theme={null}
CREATE VIEW expense_anomalies AS
SELECT 
  e.expense_id,
  e.employee_id,
  e.category,
  e.amount,
  e.submission_date,
  -- Anomaly indicators
  CASE WHEN e.amount > AVG(e.amount) OVER (PARTITION BY e.category) * 2 THEN 1 ELSE 0 END as high_amount_flag,
  CASE WHEN e.amount < AVG(e.amount) OVER (PARTITION BY e.category) * 0.1 THEN 1 ELSE 0 END as low_amount_flag,
  CASE WHEN e.category NOT IN (SELECT category FROM policies.allowed_categories) THEN 1 ELSE 0 END as policy_violation_flag
FROM expenses.expenses e;
```

### Step 3: Create MCP Tools

**Tool 1: Review Expense**

* `review_expense(expense_id: string)`

**Tool 2: Identify Anomalies**

* `identify_anomalies(days_back: number, min_amount: number)`

**Tool 3: Check Compliance**

* `check_compliance(expense_id: string, policy_version: string)`

**Tool 4: Generate Audit Report**

* `generate_audit_report(period: string, include_anomalies: boolean)`

## Example Agent Interactions

**User**: "Review expenses for potential fraud"

**Agent**: "Expense Audit Results:

* Total Expenses: \$125K
* Anomalies Found: 12
* High Risk: 3 expenses (\$15K total)
  * Employee #456: $8K meal expense (policy limit: $500)
  * Employee #789: \$5K travel (no receipts)
  * Employee #123: \$2K duplicate submission
* Recommendation: Flag for manual review"

## Outcomes

* **Review Speed**: 85% faster
* **Anomaly Detection**: 90% accuracy
* **Compliance**: 95% policy compliance
* **Fraud Detection**: 60% earlier detection

## Next Steps

* [Financial Analyst Example](/examples/financial-analyst)
* [Invoice & Billing Assistant Example](/examples/invoice-billing-assistant)