bakery-ia/docs/08-api-reference/ai-insights-api.md

# Technical Documentation - Bakery IA AI Insights Platform

## Table of Contents

1. [API Reference](#api-reference)
2. [Deployment Guide](#deployment-guide)
3. [Implementation Details](#implementation-details)
4. [Dynamic Rules Engine](#dynamic-rules-engine)
5. [Database Management](#database-management)
6. [Configuration](#configuration)
7. [Troubleshooting](#troubleshooting)

---

## API Reference

### Base URL

```
http://ai-insights-service:8000/api/v1/ai-insights
```

### Authentication

All endpoints require either:
- JWT token in `Authorization: Bearer <token>` header
- Service token in `X-Service-Token` header
- Demo session ID in `X-Demo-Session-Id` header

### Tenant Context

All endpoints include tenant ID in the path:
```
/api/v1/ai-insights/tenants/{tenant_id}/...
```

---

### Insights Endpoints

#### Create Insight

**POST** `/tenants/{tenant_id}/insights`

Creates a new AI insight.

**Request Body:**
```json
{
  "type": "prediction",              // required: prediction, recommendation, alert, optimization
  "priority": "high",                // required: critical, high, medium, low
  "category": "forecasting",         // required: forecasting, inventory, production, procurement, etc.
  "title": "Weekend Demand Surge",   // required: max 255 chars
  "description": "Detailed explanation...",  // optional: text
  "confidence": 87,                  // required: 0-100
  "metrics_json": {                  // optional: JSONB object
    "product_id": "croissant",
    "predicted_demand": 130,
    "increase_percentage": 30
  },
  "impact_type": "revenue_increase", // optional: revenue_increase, cost_reduction, etc.
  "impact_value": 450.00,           // optional: decimal
  "impact_unit": "euros",           // optional: string
  "actionable": true,               // optional: boolean, default true
  "recommendation_actions": [        // optional: array of actions
    {
      "service": "production",
      "action": "increase_production",
      "parameters": "{\"product_id\": \"croissant\", \"quantity\": 30}"
    }
  ],
  "source_service": "forecasting",  // required: originating service
  "source_data_id": "forecast_001", // optional: reference ID
  "valid_from": "2025-11-03T00:00:00Z",  // optional: ISO 8601
  "valid_until": "2025-11-05T23:59:59Z"  // optional: ISO 8601
}
```

**Response:** `201 Created`
```json
{
  "id": "uuid",
  "tenant_id": "uuid",
  "type": "prediction",
  "priority": "high",
  "category": "forecasting",
  "title": "Weekend Demand Surge",
  "description": "Detailed explanation...",
  "confidence": 87,
  "metrics_json": {...},
  "impact_type": "revenue_increase",
  "impact_value": 450.00,
  "impact_unit": "euros",
  "status": "new",
  "actionable": true,
  "recommendation_actions": [...],
  "source_service": "forecasting",
  "source_data_id": "forecast_001",
  "valid_from": "2025-11-03T00:00:00Z",
  "valid_until": "2025-11-05T23:59:59Z",
  "created_at": "2025-11-03T10:30:00Z",
  "updated_at": "2025-11-03T10:30:00Z"
}
```

---

#### List Insights

**GET** `/tenants/{tenant_id}/insights`

Retrieves paginated list of insights with optional filters.

**Query Parameters:**
- `skip` (int, default=0): Pagination offset
- `limit` (int, default=100, max=1000): Results per page
- `priority` (string): Filter by priority (critical, high, medium, low)
- `category` (string): Filter by category
- `status` (string): Filter by status (new, acknowledged, in_progress, applied, dismissed)
- `actionable_only` (boolean): Only actionable insights
- `min_confidence` (int, 0-100): Minimum confidence score

**Response:** `200 OK`
```json
{
  "items": [
    {
      "id": "uuid",
      "title": "...",
      // ... full insight object
    }
  ],
  "total": 42,
  "skip": 0,
  "limit": 100
}
```

---

#### Get Single Insight

**GET** `/tenants/{tenant_id}/insights/{insight_id}`

**Response:** `200 OK`
```json
{
  "id": "uuid",
  // ... full insight object
}
```

**Errors:**
- `404 Not Found`: Insight doesn't exist
- `403 Forbidden`: Tenant mismatch

---

#### Update Insight

**PATCH** `/tenants/{tenant_id}/insights/{insight_id}`

Updates specific fields of an insight.

**Request Body:**
```json
{
  "status": "acknowledged",  // new, acknowledged, in_progress, applied, dismissed
  "priority": "critical",    // optional: upgrade/downgrade priority
  "notes": "Additional info" // optional: any field that's updatable
}
```

**Response:** `200 OK`
```json
{
  // updated insight object
}
```

---

#### Delete Insight (Soft Delete)

**DELETE** `/tenants/{tenant_id}/insights/{insight_id}`

Marks insight as deleted (soft delete).

**Response:** `204 No Content`

---

#### Get Orchestration-Ready Insights

**GET** `/tenants/{tenant_id}/insights/orchestration-ready`

Retrieves insights grouped by category, ready for orchestration.

**Query Parameters:**
- `target_date` (ISO 8601): Target execution date
- `min_confidence` (int, default=70): Minimum confidence threshold

**Response:** `200 OK`
```json
{
  "forecast_adjustments": [
    {
      "id": "uuid",
      "title": "...",
      "confidence": 87,
      "recommendation_actions": [...]
    }
  ],
  "procurement_recommendations": [...],
  "production_optimizations": [...],
  "supplier_alerts": [...],
  "price_opportunities": [...]
}
```

---

### Feedback Endpoints

#### Record Feedback

**POST** `/tenants/{tenant_id}/insights/{insight_id}/feedback`

Records actual outcome and compares with prediction.

**Request Body:**
```json
{
  "action_taken": "increased_production",
  "success": true,
  "result_data": {
    "planned_increase": 30,
    "actual_increase": 28,
    "revenue_impact": 420.00
  },
  "expected_impact_value": 450.00,
  "actual_impact_value": 420.00,
  "variance_percentage": -6.67,
  "accuracy_score": 93.3,
  "notes": "Slightly lower than predicted due to supply constraints"
}
```

**Response:** `200 OK`
```json
{
  "id": "uuid",
  "insight_id": "uuid",
  "action_taken": "increased_production",
  "success": true,
  "result_data": {...},
  "expected_impact_value": 450.00,
  "actual_impact_value": 420.00,
  "variance_percentage": -6.67,
  "accuracy_score": 93.3,
  "notes": "...",
  "created_at": "2025-11-03T18:00:00Z"
}
```

**Side Effects:**
- Automatically updates insight status to "applied"
- Triggers FeedbackLearningSystem analysis
- May trigger model retraining if performance degrades

---

### Metrics Endpoints

#### Get Summary Metrics

**GET** `/tenants/{tenant_id}/insights/metrics/summary`

Retrieves aggregate metrics for all insights.

**Response:** `200 OK`
```json
{
  "total_insights": 147,
  "actionable_insights": 98,
  "average_confidence": 82.5,
  "critical_priority_count": 12,
  "high_priority_count": 45,
  "medium_priority_count": 67,
  "low_priority_count": 23,
  "by_category": {
    "forecasting": 42,
    "inventory": 35,
    "production": 28,
    "procurement": 22,
    "customer": 20
  },
  "by_status": {
    "new": 56,
    "acknowledged": 28,
    "in_progress": 15,
    "applied": 42,
    "dismissed": 6
  }
}
```

---

## Deployment Guide

### Prerequisites

- **Kubernetes Cluster:** 1.24+
- **Docker:** 20.10+
- **Kind:** 0.20+ (for local development)
- **kubectl:** 1.24+
- **Tilt:** 0.30+ (optional, for development)

### Local Development Setup

#### 1. Start Kubernetes Cluster

```bash
# Create Kind cluster
kind create cluster --name bakery-ia-local --config infrastructure/kind-config.yaml

# Verify cluster
kubectl cluster-info
```

#### 2. Deploy Infrastructure

```bash
# Create namespace
kubectl create namespace bakery-ia

# Deploy databases
kubectl apply -f infrastructure/kubernetes/base/components/databases/

# Wait for databases to be ready
kubectl wait --for=condition=ready pod -l app=postgresql-main -n bakery-ia --timeout=300s
kubectl wait --for=condition=ready pod -l app=postgresql-ai-insights -n bakery-ia --timeout=300s
kubectl wait --for=condition=ready pod -l app=redis -n bakery-ia --timeout=300s
```

#### 3. Deploy Services

```bash
# Deploy all services
kubectl apply -f infrastructure/kubernetes/base/

# Watch deployment
kubectl get pods -n bakery-ia -w
```

#### 4. Run Database Migrations

```bash
# AI Insights Service migration
kubectl exec -n bakery-ia deployment/ai-insights-service -- \
  python -m alembic upgrade head

# Other services...
for service in orders inventory production suppliers; do
  kubectl exec -n bakery-ia deployment/${service}-service -- \
    python -m alembic upgrade head
done
```

#### 5. Verify Deployment

```bash
# Check all pods are running
kubectl get pods -n bakery-ia

# Check services
kubectl get svc -n bakery-ia

# Test AI Insights Service health
kubectl port-forward -n bakery-ia svc/ai-insights-service 8000:8000 &
curl http://localhost:8000/health
```

---

### Production Deployment

#### Environment Configuration

Create environment-specific configurations:

```yaml
# infrastructure/kubernetes/overlays/production/kustomization.yaml
apiVersion: kustomize.config.k8s.io/v1beta1
kind: Kustomization

resources:
  - ../../base

replicas:
  - name: ai-insights-service
    count: 3
  - name: orchestration-service
    count: 2

configMapGenerator:
  - name: ai-insights-config
    env: production.env

secretGenerator:
  - name: database-secrets
    envs:
      - secrets.env

images:
  - name: bakery/ai-insights-service
    newTag: v1.0.0
```

#### Deploy to Production

```bash
# Apply with kustomize
kubectl apply -k infrastructure/kubernetes/overlays/production/

# Rolling update
kubectl set image deployment/ai-insights-service \
  ai-insights-service=bakery/ai-insights-service:v1.0.1 \
  -n bakery-ia

# Monitor rollout
kubectl rollout status deployment/ai-insights-service -n bakery-ia
```

---

### Database Management

#### Create AI Insights Database

```bash
# Connect to PostgreSQL
kubectl exec -it -n bakery-ia postgresql-ai-insights-0 -- psql -U postgres

# Create database
CREATE DATABASE ai_insights_db;

# Create user
CREATE USER ai_insights_user WITH PASSWORD 'secure_password';
GRANT ALL PRIVILEGES ON DATABASE ai_insights_db TO ai_insights_user;

# Enable UUID extension
\c ai_insights_db
CREATE EXTENSION IF NOT EXISTS "uuid-ossp";
```

#### Run Migrations

```bash
# Check current version
kubectl exec -n bakery-ia deployment/ai-insights-service -- \
  python -m alembic current

# Upgrade to latest
kubectl exec -n bakery-ia deployment/ai-insights-service -- \
  python -m alembic upgrade head

# Downgrade one version
kubectl exec -n bakery-ia deployment/ai-insights-service -- \
  python -m alembic downgrade -1

# Show migration history
kubectl exec -n bakery-ia deployment/ai-insights-service -- \
  python -m alembic history
```

#### Backup and Restore

```bash
# Backup
kubectl exec -n bakery-ia postgresql-ai-insights-0 -- \
  pg_dump -U postgres ai_insights_db > backup-$(date +%Y%m%d).sql

# Restore
kubectl exec -i -n bakery-ia postgresql-ai-insights-0 -- \
  psql -U postgres ai_insights_db < backup-20251103.sql
```

---

## Implementation Details

### Service Structure

```
services/ai_insights/
├── app/
│   ├── __init__.py
│   ├── main.py                    # FastAPI application
│   ├── core/
│   │   ├── config.py             # Configuration
│   │   ├── database.py           # Database connection
│   │   └── security.py           # Auth utilities
│   ├── models/
│   │   ├── ai_insight.py         # SQLAlchemy models
│   │   └── feedback.py
│   ├── schemas/
│   │   ├── insight.py            # Pydantic schemas
│   │   └── feedback.py
│   ├── api/
│   │   ├── insights.py           # Insight endpoints
│   │   ├── feedback.py           # Feedback endpoints
│   │   └── metrics.py            # Metrics endpoints
│   ├── services/
│   │   ├── insight_service.py    # Business logic
│   │   └── feedback_service.py
│   ├── repositories/
│   │   ├── insight_repository.py # Data access
│   │   └── feedback_repository.py
│   └── ml/
│       └── feedback_learning_system.py  # Learning system
├── tests/
│   ├── unit/
│   ├── integration/
│   └── conftest.py
├── migrations/
│   └── versions/                  # Alembic migrations
├── Dockerfile
├── requirements.txt
└── alembic.ini
```

### Key Components

#### FastAPI Application

```python
# app/main.py
from fastapi import FastAPI
from app.api import insights, feedback, metrics
from app.core.database import engine
from app.models import Base

app = FastAPI(
    title="AI Insights Service",
    version="1.0.0",
    description="Centralized AI insights management"
)

# Include routers
app.include_router(insights.router, prefix="/api/v1/ai-insights", tags=["insights"])
app.include_router(feedback.router, prefix="/api/v1/ai-insights", tags=["feedback"])
app.include_router(metrics.router, prefix="/api/v1/ai-insights", tags=["metrics"])

@app.on_event("startup")
async def startup():
    # Initialize database
    async with engine.begin() as conn:
        await conn.run_sync(Base.metadata.create_all)

@app.get("/health")
async def health_check():
    return {"status": "healthy", "service": "ai-insights"}
```

#### Repository Pattern

```python
# app/repositories/insight_repository.py
from sqlalchemy.ext.asyncio import AsyncSession
from sqlalchemy import select, and_
from app.models import AIInsight

class InsightRepository:
    def __init__(self, session: AsyncSession):
        self.session = session

    async def create(self, insight_data: dict) -> AIInsight:
        insight = AIInsight(**insight_data)
        self.session.add(insight)
        await self.session.commit()
        await self.session.refresh(insight)
        return insight

    async def get_by_id(self, tenant_id: str, insight_id: str) -> AIInsight:
        query = select(AIInsight).where(
            and_(
                AIInsight.tenant_id == tenant_id,
                AIInsight.id == insight_id,
                AIInsight.deleted_at.is_(None)
            )
        )
        result = await self.session.execute(query)
        return result.scalar_one_or_none()

    async def list_insights(
        self,
        tenant_id: str,
        skip: int = 0,
        limit: int = 100,
        **filters
    ) -> tuple[list[AIInsight], int]:
        # Build query with filters
        query = select(AIInsight).where(
            and_(
                AIInsight.tenant_id == tenant_id,
                AIInsight.deleted_at.is_(None)
            )
        )

        # Apply filters
        if priority := filters.get('priority'):
            query = query.where(AIInsight.priority == priority)
        if category := filters.get('category'):
            query = query.where(AIInsight.category == category)
        if min_confidence := filters.get('min_confidence'):
            query = query.where(AIInsight.confidence >= min_confidence)

        # Get total count
        count_query = select(func.count()).select_from(query.subquery())
        total = await self.session.execute(count_query)
        total = total.scalar()

        # Apply pagination
        query = query.offset(skip).limit(limit)
        result = await self.session.execute(query)

        return result.scalars().all(), total
```

---

## Dynamic Rules Engine

The Dynamic Rules Engine adapts business rules based on historical patterns.

### Architecture

```
Historical Data
    ↓
Pattern Detector (analyzes trends, seasonality, anomalies)
    ↓
Rules Orchestrator (adapts thresholds and parameters)
    ↓
Rule Evaluation (applies adapted rules to current data)
    ↓
Insights Generated
```

### Rule Types

1. **Demand Threshold Rules**
   - High demand alert: demand > adaptive_threshold
   - Low demand alert: demand < adaptive_threshold
   - Threshold adapts based on historical mean and variance

2. **Volatility Rules**
   - Triggered when coefficient of variation > threshold
   - Warns of unpredictable demand patterns

3. **Trend Rules**
   - Upward trend: sustained increase over N periods
   - Downward trend: sustained decrease over N periods

4. **Seasonal Rules**
   - Detects recurring patterns (weekly, monthly)
   - Adjusts baselines for seasonal effects

5. **Anomaly Rules**
   - Statistical outliers (> 3 standard deviations)
   - Sudden changes (> X% from baseline)

### Usage Example

```python
from app.ml.dynamic_rules_engine import DynamicRulesEngine

# Initialize engine
engine = DynamicRulesEngine(tenant_id=tenant_id)

# Train on historical data
historical_data = pd.DataFrame({
    'date': [...],
    'product_id': [...],
    'quantity': [...]
})

engine.train(historical_data)

# Generate insights for current data
current_data = pd.DataFrame({
    'product_id': ['croissant'],
    'current_demand': [130],
    'date': ['2025-11-03']
})

insights = await engine.generate_insights(current_data)

# Store insights in AI Insights Service
for insight in insights:
    await insight_service.create_insight(tenant_id, insight)
```

### Configuration

```python
# services/forecasting/app/core/config.py
class RulesEngineSettings(BaseSettings):
    # Thresholds
    HIGH_DEMAND_THRESHOLD: float = 1.2  # 20% above baseline
    LOW_DEMAND_THRESHOLD: float = 0.8   # 20% below baseline
    VOLATILITY_THRESHOLD: float = 0.3   # CV > 30%

    # Pattern detection
    SEASONALITY_PERIODS: list[int] = [7, 30]  # Weekly, monthly
    TREND_WINDOW: int = 14  # Days to detect trends
    ANOMALY_SIGMA: float = 3.0  # Standard deviations

    # Adaptation
    ADAPTATION_RATE: float = 0.1  # How quickly to adapt thresholds
    MIN_SAMPLES: int = 30  # Minimum data points for adaptation
    CONFIDENCE_DECAY: float = 0.95  # Confidence decay over time
```

---

## Configuration

### Environment Variables

```bash
# Database
AI_INSIGHTS_DATABASE_URL=postgresql+asyncpg://user:pass@host:5432/ai_insights_db
DATABASE_POOL_SIZE=20
DATABASE_MAX_OVERFLOW=10

# Redis
REDIS_URL=redis://redis:6379/0

# Service URLs
FORECASTING_SERVICE_URL=http://forecasting-service:8000
PRODUCTION_SERVICE_URL=http://production-service:8000
INVENTORY_SERVICE_URL=http://inventory-service:8000
PROCUREMENT_SERVICE_URL=http://procurement-service:8000
ORCHESTRATION_SERVICE_URL=http://orchestration-service:8000

# Authentication
JWT_SECRET_KEY=your-secret-key-here
JWT_ALGORITHM=HS256
JWT_EXPIRATION_MINUTES=60

# Logging
LOG_LEVEL=INFO
LOG_FORMAT=json

# ML Configuration
MIN_CONFIDENCE_THRESHOLD=70
RETRAINING_ACCURACY_THRESHOLD=0.75
FEEDBACK_SAMPLE_SIZE=100
```

### Kubernetes ConfigMap

```yaml
apiVersion: v1
kind: ConfigMap
metadata:
  name: ai-insights-config
  namespace: bakery-ia
data:
  LOG_LEVEL: "INFO"
  MIN_CONFIDENCE_THRESHOLD: "70"
  FORECASTING_SERVICE_URL: "http://forecasting-service:8000"
  PRODUCTION_SERVICE_URL: "http://production-service:8000"
  INVENTORY_SERVICE_URL: "http://inventory-service:8000"
```

### Kubernetes Secrets

```yaml
apiVersion: v1
kind: Secret
metadata:
  name: database-secrets
  namespace: bakery-ia
type: Opaque
stringData:
  AI_INSIGHTS_DATABASE_URL: "postgresql+asyncpg://user:pass@postgresql-ai-insights:5432/ai_insights_db"
  REDIS_URL: "redis://redis:6379/0"
  JWT_SECRET_KEY: "your-secure-secret-key"
```

---

## Troubleshooting

### Common Issues

#### 1. Service Not Starting

```bash
# Check pod logs
kubectl logs -n bakery-ia deployment/ai-insights-service --tail=100

# Check pod events
kubectl describe pod -n bakery-ia <pod-name>

# Common causes:
# - Database connection failure
# - Missing environment variables
# - Port conflicts
```

#### 2. Database Connection Errors

```bash
# Test database connectivity
kubectl exec -it -n bakery-ia deployment/ai-insights-service -- \
  python -c "from app.core.database import engine; import asyncio; asyncio.run(engine.connect())"

# Check database pod status
kubectl get pods -n bakery-ia -l app=postgresql-ai-insights

# Verify database URL
kubectl exec -n bakery-ia deployment/ai-insights-service -- \
  env | grep DATABASE_URL
```

#### 3. High Memory Usage

```bash
# Check resource usage
kubectl top pods -n bakery-ia

# Increase limits
kubectl set resources deployment/ai-insights-service \
  --limits=memory=2Gi \
  -n bakery-ia

# Enable query result streaming for large datasets
# (already implemented in repository pattern)
```

#### 4. Slow API Responses

```bash
# Check database query performance
kubectl exec -it -n bakery-ia postgresql-ai-insights-0 -- \
  psql -U postgres -d ai_insights_db -c "
    SELECT query, calls, mean_exec_time, total_exec_time
    FROM pg_stat_statements
    ORDER BY total_exec_time DESC
    LIMIT 10;
  "

# Add missing indexes if needed
# Check slow query log
kubectl logs -n bakery-ia -l app=postgresql-ai-insights | grep "duration"
```

#### 5. Insight Creation Failures

```bash
# Check validation errors
kubectl logs -n bakery-ia deployment/ai-insights-service | grep -i error

# Common issues:
# - Invalid confidence score (must be 0-100)
# - Missing required fields
# - Invalid tenant ID
# - Database constraint violations
```

### Debugging Commands

```bash
# Interactive shell in pod
kubectl exec -it -n bakery-ia deployment/ai-insights-service -- /bin/bash

# Python REPL with app context
kubectl exec -it -n bakery-ia deployment/ai-insights-service -- \
  python -c "from app.core.database import engine; import asyncio; # your code"

# Check API health
kubectl exec -n bakery-ia deployment/ai-insights-service -- \
  curl http://localhost:8000/health

# View recent logs with timestamps
kubectl logs -n bakery-ia deployment/ai-insights-service \
  --since=1h \
  --timestamps

# Follow logs in real-time
kubectl logs -n bakery-ia deployment/ai-insights-service -f
```

---

## Performance Optimization

### Database Optimization

```sql
-- Create covering indexes
CREATE INDEX idx_insights_tenant_priority_confidence
ON ai_insights(tenant_id, priority, confidence)
WHERE deleted_at IS NULL;

-- Vacuum regularly
VACUUM ANALYZE ai_insights;

-- Check index usage
SELECT schemaname, tablename, indexname, idx_scan
FROM pg_stat_user_indexes
WHERE schemaname = 'public'
ORDER BY idx_scan;
```

### Redis Caching

```python
# Cache frequently accessed insights
from app.core.cache import redis_client

async def get_insight_cached(tenant_id: str, insight_id: str):
    # Check cache
    cache_key = f"insight:{tenant_id}:{insight_id}"
    cached = await redis_client.get(cache_key)

    if cached:
        return json.loads(cached)

    # Fetch from database
    insight = await repository.get_by_id(tenant_id, insight_id)

    # Cache for 5 minutes
    await redis_client.setex(
        cache_key,
        300,
        json.dumps(insight.dict())
    )

    return insight
```

### Batch Operations

```python
# Bulk insert insights
async def create_insights_batch(tenant_id: str, insights_data: list[dict]):
    async with session.begin():
        insights = [AIInsight(**data) for data in insights_data]
        session.add_all(insights)
        await session.flush()
        return insights
```

---

## Monitoring and Observability

### Health Checks

```python
@app.get("/health")
async def health_check():
    return {
        "status": "healthy",
        "service": "ai-insights",
        "version": "1.0.0",
        "timestamp": datetime.utcnow().isoformat()
    }

@app.get("/health/detailed")
async def detailed_health_check():
    # Check database
    try:
        async with engine.connect() as conn:
            await conn.execute(text("SELECT 1"))
        db_status = "healthy"
    except Exception as e:
        db_status = f"unhealthy: {str(e)}"

    # Check Redis
    try:
        await redis_client.ping()
        redis_status = "healthy"
    except Exception as e:
        redis_status = f"unhealthy: {str(e)}"

    return {
        "status": "healthy" if db_status == "healthy" and redis_status == "healthy" else "unhealthy",
        "components": {
            "database": db_status,
            "redis": redis_status
        }
    }
```

### Metrics Endpoint

```python
from prometheus_client import Counter, Histogram, generate_latest

insight_created = Counter('insights_created_total', 'Total insights created')
insight_applied = Counter('insights_applied_total', 'Total insights applied')
api_latency = Histogram('api_request_duration_seconds', 'API request latency')

@app.get("/metrics")
async def metrics():
    return Response(generate_latest(), media_type="text/plain")
```

---

*For comprehensive testing procedures, validation steps, and test cases, refer to TESTING_GUIDE.md.*