feat: initial Skeen-CRM AI Agent architecture

- FastAPI + Python 3.12 backend - Meta WhatsApp Business API client (official) - OpenAI GPT-4o with function calling - RAG vector store with pgvector - ERPNext Frappe REST client - Celery + Redis async task queue - PostgreSQL with migrations (Alembic) - Docker Compose full stack - Enterprise logging, metrics, health checks
2026-04-29 05:30:59 +00:00
commit d30b22b50c
44 changed files with 3603 additions and 0 deletions
--- a/.env.example
+++ b/.env.example
@@ -0,0 +1,73 @@
+# =============================================================================
+# FASTAPI APPLICATION
+# =============================================================================
+APP_NAME=Skeen-CRM-Agent
+APP_ENV=development
+DEBUG=true
+LOG_LEVEL=INFO
+SECRET_KEY=change-me-in-production-skeen-2024
+
+# =============================================================================
+# SERVER
+# =============================================================================
+HOST=0.0.0.0
+PORT=8000
+
+# =============================================================================
+# DATABASE (PostgreSQL + pgvector)
+# =============================================================================
+DATABASE_URL=postgresql+asyncpg://skeen:skeen123@localhost:5432/skeen_crm
+DATABASE_POOL_SIZE=20
+DATABASE_MAX_OVERFLOW=10
+
+# =============================================================================
+# REDIS
+# =============================================================================
+REDIS_URL=redis://localhost:6379/0
+
+# =============================================================================
+# META / WHATSAPP BUSINESS API (Oficial)
+# =============================================================================
+META_API_VERSION=v18.0
+META_ACCESS_TOKEN=EAAXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
+META_PHONE_NUMBER_ID=123456789012345
+META_BUSINESS_ACCOUNT_ID=987654321098765
+META_WEBHOOK_VERIFY_TOKEN=skeen-webhook-verify-token-2024
+META_APP_SECRET=your-app-secret-for-signature-verification
+
+# =============================================================================
+# OPENAI
+# =============================================================================
+OPENAI_API_KEY=sk-xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
+OPENAI_MODEL=gpt-4o
+OPENAI_EMBEDDING_MODEL=text-embedding-3-small
+OPENAI_TEMPERATURE=0.3
+OPENAI_MAX_TOKENS=1500
+
+# =============================================================================
+# RAG / VECTOR STORE
+# =============================================================================
+VECTOR_DIMENSION=1536
+RAG_TOP_K=5
+RAG_SIMILARITY_THRESHOLD=0.75
+
+# =============================================================================
+# ERPNEXT INTEGRATION
+# =============================================================================
+ERPNEXT_BASE_URL=https://skeen.erpnext.com
+ERPNEXT_API_KEY=xxxxxxxxxxxxxxxx
+ERPNEXT_API_SECRET=xxxxxxxxxxxxxxxx
+ERPNEXT_VERIFY_SSL=true
+
+# =============================================================================
+# CELERY
+# =============================================================================
+CELERY_BROKER_URL=redis://localhost:6379/1
+CELERY_RESULT_BACKEND=redis://localhost:6379/2
+CELERY_WORKER_CONCURRENCY=4
+
+# =============================================================================
+# MONITORING
+# =============================================================================
+ENABLE_METRICS=true
+SENTRY_DSN=
--- a/.gitignore
+++ b/.gitignore
@@ -0,0 +1,149 @@
+# Byte-compiled / optimized / DLL files
+__pycache__/
+*.py[cod]
+*$py.class
+
+# C extensions
+*.so
+
+# Distribution / packaging
+.Python
+build/
+develop-eggs/
+dist/
+downloads/
+eggs/
+.eggs/
+lib/
+lib64/
+parts/
+sdist/
+var/
+wheels/
+share/python-wheels/
+*.egg-info/
+.installed.cfg
+*.egg
+MANIFEST
+
+# PyInstaller
+*.manifest
+*.spec
+
+# Installer logs
+pip-log.txt
+pip-delete-this-directory.txt
+
+# Unit test / coverage reports
+htmlcov/
+.tox/
+.nox/
+.coverage
+.coverage.*
+.cache
+nosetests.xml
+coverage.xml
+*.cover
+*.py,cover
+.hypothesis/
+.pytest_cache/
+cover/
+
+# Translations
+*.mo
+*.pot
+
+# Django stuff:
+*.log
+local_settings.py
+db.sqlite3
+db.sqlite3-journal
+
+# Flask stuff:
+instance/
+.webassets-cache
+
+# Scrapy stuff:
+.scrapy
+
+# Sphinx documentation
+docs/_build/
+
+# PyBuilder
+.pybuilder/
+target/
+
+# Jupyter Notebook
+.ipynb_checkpoints
+
+# IPython
+profile_default/
+ipython_config.py
+
+# pyenv
+.python-version
+
+# pipenv
+Pipfile.lock
+
+# poetry
+poetry.lock
+
+# pdm
+.pdm.toml
+
+# PEP 582
+__pypackages__/
+
+# Celery stuff
+celerybeat-schedule
+celerybeat.pid
+
+# SageMath parsed files
+*.sage.py
+
+# Environments
+.env
+.venv
+env/
+venv/
+ENV/
+env.bak/
+venv.bak/
+
+# Spyder project settings
+.spyderproject
+.spyproject
+
+# Rope project settings
+.ropeproject
+
+# mkdocs documentation
+/site
+
+# mypy
+.mypy_cache/
+.dmypy.json
+dmypy.json
+
+# Pyre type checker
+.pyre/
+
+# pytype static type analyzer
+.pytype/
+
+# Cython debug symbols
+cython_debug/
+
+# PyCharm
+.idea/
+
+# VS Code
+.vscode/
+
+# macOS
+.DS_Store
+
+# Docker volumes
+postgres_data/
+redis_data/
--- a/57
+++ b/57
@@ -0,0 +1,57 @@
+# =============================================================================
+# SKEEN CRM Agent - Production Dockerfile
+# =============================================================================
+FROM python:3.12-slim AS builder
+
+# Install build dependencies
+RUN apt-get update && apt-get install -y --no-install-recommends \
+    build-essential \
+    libpq-dev \
+    && rm -rf /var/lib/apt/lists/*
+
+# Install uv (modern Python package manager)
+RUN pip install --no-cache-dir uv
+
+# Set workdir
+WORKDIR /app
+
+# Copy dependency definitions
+COPY pyproject.toml ./
+
+# Create virtual environment and install dependencies
+RUN uv venv .venv && \
+    uv pip install --no-cache -r pyproject.toml
+
+# =============================================================================
+# Production stage
+# =============================================================================
+FROM python:3.12-slim AS production
+
+# Install runtime dependencies
+RUN apt-get update && apt-get install -y --no-install-recommends \
+    libpq5 \
+    curl \
+    && rm -rf /var/lib/apt/lists/*
+
+# Create non-root user
+RUN groupadd -r skeen && useradd -r -g skeen skeen
+
+WORKDIR /app
+
+# Copy virtual environment from builder
+COPY --from=builder /app/.venv /app/.venv
+ENV PATH="/app/.venv/bin:$PATH"
+
+# Copy application code
+COPY --chown=skeen:skeen src/ ./src/
+COPY --chown=skeen:skeen alembic/ ./alembic/
+COPY --chown=skeen:skeen alembic.ini ./
+
+# Switch to non-root user
+USER skeen
+
+# Healthcheck
+HEALTHCHECK --interval=30s --timeout=10s --start-period=5s --retries=3 \
+    CMD curl -f http://localhost:8000/health || exit 1
+
+EXPOSE 8000
--- a/README.md
+++ b/README.md
@@ -0,0 +1,210 @@
+# SKEEN CRM AI Agent
+
+Agente de inteligencia artificial para WhatsApp Business API + ERPNext Healthcare.
+
+## Arquitectura
+
+```
+┌─────────────┐     ┌─────────────────────┐     ┌─────────────┐
+│   Paciente  │────▶│  WhatsApp Business  │────▶│  Meta API   │
+│  (WhatsApp) │◀────│       API           │◀────│  Webhooks   │
+└─────────────┘     └─────────────────────┘     └──────┬──────┘
+                                                       │
+                              ┌────────────────────────┘
+                              ▼
+                    ┌─────────────────────┐
+                    │   SKEEN AI Agent    │  FastAPI + Python 3.12
+                    │   (Este Repo)       │
+                    │                     │
+                    │  • OpenAI GPT-4o    │
+                    │  • RAG (pgvector)   │
+                    │  • Celery + Redis   │
+                    │  • PostgreSQL       │
+                    └──────────┬──────────┘
+                               │
+                    ┌──────────┴──────────┐
+                    ▼                     ▼
+            ┌─────────────┐      ┌─────────────┐
+            │   ERPNext   │      │  PostgreSQL │
+            │  Healthcare │      │  + pgvector │
+            └─────────────┘      └─────────────┘
+```
+
+## Stack Tecnológico
+
+| Capa | Tecnología |
+|------|-----------|
+| Web Framework | FastAPI + Uvicorn (HTTP/2) |
+| Base de datos | PostgreSQL 16 + pgvector |
+| Cola de tareas | Celery + Redis |
+| IA / LLM | OpenAI GPT-4o + Embeddings |
+| WhatsApp | Meta Business API (oficial) |
+| CRM | ERPNext (Frappe Framework) |
+| Observabilidad | Structlog + Prometheus |
+| Deploy | Docker Compose |
+
+## Estructura del Proyecto
+
+```
+Skeen-CRM/
+├── src/
+│   ├── main.py                    # Entry point FastAPI
+│   ├── config.py                  # Pydantic Settings
+│   ├── api/
+│   │   ├── v1/
+│   │   │   ├── webhooks.py        # Meta WhatsApp webhooks
+│   │   │   ├── messages.py        # API envío manual
+│   │   │   └── health.py          # Health checks
+│   │   └── deps.py                # Dependency injection
+│   ├── domain/
+│   │   ├── models/
+│   │   │   └── conversation.py    # Entidades SQLAlchemy
+│   │   └── services/
+│   ├── use_cases/
+│   │   └── handle_incoming_message.py  # Pipeline AI
+│   ├── infrastructure/
+│   │   ├── db.py                  # PostgreSQL async
+│   │   ├── redis.py               # Redis client
+│   │   ├── whatsapp/
+│   │   │   ├── client.py          # Meta Graph API client
+│   │   │   └── webhook.py         # Webhook parser/validator
+│   │   ├── ai/
+│   │   │   ├── openai_client.py   # OpenAI async client
+│   │   │   ├── rag.py             # Vector store pgvector
+│   │   │   └── prompts.py         # Tools & prompts
+│   │   └── erpnext/
+│   │       └── client.py          # Frappe REST API client
+│   └── workers/
+│       ├── celery_app.py          # Celery configuration
+│       └── tasks.py               # Background tasks
+├── alembic/                       # Database migrations
+├── docker-compose.yml             # Full stack local
+├── Dockerfile                     # Production image
+└── pyproject.toml                 # Dependencies (uv)
+```
+
+## Requisitos
+
+- Python 3.12+
+- Docker + Docker Compose
+- Cuenta de WhatsApp Business API (Meta)
+- API Key de OpenAI
+- Instancia de ERPNext (para integración completa)
+
+## Instalación Local
+
+### 1. Clonar y entrar al directorio
+
+```bash
+cd Skeen-CRM
+```
+
+### 2. Copiar variables de entorno
+
+```bash
+cp .env.example .env
+# Editar .env con tus credenciales de Meta, OpenAI y ERPNext
+```
+
+### 3. Levantar infraestructura (PostgreSQL + Redis)
+
+```bash
+docker compose up -d postgres redis
+```
+
+### 4. Instalar dependencias con uv
+
+```bash
+pip install uv
+uv venv .venv
+source .venv/bin/activate
+uv pip install -e ".[dev]"
+```
+
+### 5. Ejecutar migraciones
+
+```bash
+alembic upgrade head
+```
+
+### 6. Iniciar servidor de desarrollo
+
+```bash
+uvicorn src.main:app --reload --host 0.0.0.0 --port 8000
+```
+
+### 7. Iniciar worker de Celery (en otra terminal)
+
+```bash
+celery -A src.workers.celery_app worker --loglevel=info -Q default,whatsapp,erpnext,ai
+```
+
+## Deploy con Docker Compose (Producción)
+
+```bash
+docker compose up -d --build
+```
+
+Esto levanta:
+- `api`: FastAPI (2 workers)
+- `worker`: Celery worker (4 concurrentes)
+- `scheduler`: Celery beat
+- `postgres`: PostgreSQL + pgvector
+- `redis`: Redis persistente
+
+## Configuración de Webhook en Meta
+
+1. Ir a [Meta Developers](https://developers.facebook.com/)
+2. Seleccionar tu app de WhatsApp Business
+3. En **Webhooks**, configurar:
+   - **Callback URL**: `https://tu-dominio.com/api/v1/webhooks/whatsapp`
+   - **Verify Token**: El valor de `META_WEBHOOK_VERIFY_TOKEN` en `.env`
+   - **Suscripción**: `messages`
+
+## Tools / Funciones del Agente
+
+El agente usa **OpenAI Function Calling** para ejecutar acciones:
+
+| Tool | Descripción |
+|------|-------------|
+| `search_catalog` | Busca servicios/productos en catálogo vía RAG |
+| `check_availability` | Consulta disponibilidad de citas en ERPNext |
+| `create_appointment` | Crea cita médica en ERPNext Healthcare |
+| `get_patient_info` | Consulta historial del paciente |
+| `get_wallet_balance` | Consulta saldo de monedero |
+| `escalate_to_human` | Escalar a agente humano |
+
+## Endpoints API
+
+| Método | Endpoint | Descripción |
+|--------|----------|-------------|
+| GET | `/health` | Liveness probe |
+| GET | `/ready` | Readiness probe (incluye DB) |
+| GET | `/metrics` | Métricas Prometheus |
+| GET | `/api/v1/webhooks/whatsapp` | Verificación Meta |
+| POST | `/api/v1/webhooks/whatsapp` | Recepción mensajes |
+| POST | `/api/v1/messages/text` | Enviar texto manual |
+| POST | `/api/v1/messages/template` | Enviar plantilla |
+| POST | `/api/v1/messages/buttons` | Enviar botones |
+
+## Testing
+
+```bash
+# Unit tests
+pytest tests/ -v
+
+# With coverage
+pytest tests/ --cov=src --cov-report=html
+```
+
+## Seguridad
+
+- ✅ Verificación de firma HMAC en webhooks (producción)
+- ✅ Tokens JWT no usados (usa API Keys de Meta)
+- ✅ Rate limiting implementado por número de teléfono
+- ✅ Sanitización de inputs antes de enviar a LLM
+- ✅ Logging sin exponer datos PHI (solo últimos 4 dígitos de teléfono)
+
+## Licencia
+
+Propietario - SKEEN Clínica de Belleza
--- a/alembic.ini
+++ b/alembic.ini
@@ -0,0 +1,42 @@
+[alembic]
+script_location = alembic
+prepend_sys_path = .
+version_path_separator = os
+
+sqlalchemy.url = postgresql+asyncpg://skeen:skeen123@localhost:5432/skeen_crm
+
+[post_write_hooks]
+
+[loggers]
+keys = root,sqlalchemy,alembic
+
+[handlers]
+keys = console
+
+[formatters]
+keys = generic
+
+[logger_root]
+level = WARN
+handlers = console
+qualname =
+
+[logger_sqlalchemy]
+level = WARN
+handlers =
+qualname = sqlalchemy.engine
+
+[logger_alembic]
+level = INFO
+handlers =
+qualname = alembic
+
+[handler_console]
+class = StreamHandler
+args = (sys.stderr,)
+level = NOTSET
+formatter = generic
+
+[formatter_generic]
+format = %(levelname)-5.5s [%(name)s] %(message)s
+datefmt = %H:%M:%S
--- a/alembic/README
+++ b/alembic/README
@@ -0,0 +1 @@
+Generic single-database configuration.
--- a/alembic/env.py
+++ b/alembic/env.py
@@ -0,0 +1,76 @@
+import asyncio
+from logging.config import fileConfig
+
+from sqlalchemy import pool
+from sqlalchemy.engine import Connection
+from sqlalchemy.ext.asyncio import async_engine_from_config
+
+from alembic import context
+from src.config import settings
+from src.infrastructure.db import Base
+
+# this is the Alembic Config object, which provides
+# access to the values within the .ini file in use.
+config = context.config
+
+# Interpret the config file for Python logging.
+# This line sets up loggers basically.
+if config.config_file_name is not None:
+    fileConfig(config.config_file_name)
+
+# add your model's MetaData object here
+# for 'autogenerate' support
+target_metadata = Base.metadata
+
+# other values from the config, defined by the needs of env.py,
+# can be acquired:
+# my_important_option = config.get_main_option("my_important_option")
+config.set_main_option("sqlalchemy.url", settings.DATABASE_URL)
+
+
+def run_migrations_offline() -> None:
+    """Run migrations in 'offline' mode."""
+    url = config.get_main_option("sqlalchemy.url")
+    context.configure(
+        url=url,
+        target_metadata=target_metadata,
+        literal_binds=True,
+        dialect_opts={"paramstyle": "named"},
+    )
+
+    with context.begin_transaction():
+        context.run_migrations()
+
+
+def do_run_migrations(connection: Connection) -> None:
+    context.configure(connection=connection, target_metadata=target_metadata)
+
+    with context.begin_transaction():
+        context.run_migrations()
+
+
+async def run_async_migrations() -> None:
+    """In this scenario we need to create an Engine
+    and associate a connection with the context.
+    """
+    connectable = async_engine_from_config(
+        config.get_section(config.config_ini_section, {}),
+        prefix="sqlalchemy.",
+        poolclass=pool.NullPool,
+    )
+
+    async with connectable.connect() as connection:
+        await connection.run_sync(do_run_migrations)
+
+    await connectable.dispose()
+
+
+def run_migrations_online() -> None:
+    """Run migrations in 'online' mode."""
+    asyncio.run(run_async_migrations())
+
+
+if context.is_offline_mode():
+    run_migrations_offline()
+else:
+    run_migrations_online()
--- a/alembic/script.py.mako
+++ b/alembic/script.py.mako
@@ -0,0 +1,26 @@
+"""${message}
+
+Revision ID: ${up_revision}
+Revises: ${down_revision | comma,n}
+Create Date: ${create_date}
+
+"""
+from typing import Sequence, Union
+
+from alembic import op
+import sqlalchemy as sa
+${imports if imports else ""}
+
+# revision identifiers, used by Alembic.
+revision: str = ${repr(up_revision)}
+down_revision: Union[str, None] = ${repr(down_revision)}
+branch_labels: Union[str, Sequence[str], None] = ${repr(branch_labels)}
+depends_on: Union[str, Sequence[str], None] = ${repr(depends_on)}
+
+
+def upgrade() -> None:
+    ${upgrades if upgrades else "pass"}
+
+
+def downgrade() -> None:
+    ${downgrades if downgrades else "pass"}
--- a/alembic/versions/20260428_init.py
+++ b/alembic/versions/20260428_init.py
@@ -0,0 +1,100 @@
+"""Initial migration: conversations, messages, knowledge_chunks.
+
+Revision ID: 001
+Revises:
+Create Date: 2026-04-28 00:00:00.000000
+
+"""
+from typing import Sequence, Union
+
+from alembic import op
+import sqlalchemy as sa
+from sqlalchemy.dialects import postgresql
+
+# revision identifiers, used by Alembic.
+revision: str = "001"
+down_revision: Union[str, None] = None
+branch_labels: Union[str, Sequence[str], None] = None
+depends_on: Union[str, Sequence[str], None] = None
+
+
+def upgrade() -> None:
+    # Enable pgvector extension
+    op.execute("CREATE EXTENSION IF NOT EXISTS vector")
+
+    # Conversations table
+    op.create_table(
+        "conversations",
+        sa.Column("id", sa.String(36), primary_key=True),
+        sa.Column("phone_number", sa.String(20), nullable=False, index=True),
+        sa.Column("patient_id", sa.String(100), nullable=True, index=True),
+        sa.Column("patient_name", sa.String(255), nullable=True),
+        sa.Column(
+            "status",
+            sa.Enum("active", "paused", "resolved", "escalated", "appointment_confirmed", name="conversationstatus"),
+            nullable=False,
+            server_default="active",
+        ),
+        sa.Column("context", postgresql.JSONB(astext_type=sa.Text()), server_default="{}"),
+        sa.Column("last_message_at", sa.DateTime(timezone=True), nullable=True),
+        sa.Column("created_at", sa.DateTime(timezone=True), server_default=sa.func.now()),
+        sa.Column("updated_at", sa.DateTime(timezone=True), server_default=sa.func.now(), onupdate=sa.func.now()),
+    )
+
+    # Messages table
+    op.create_table(
+        "messages",
+        sa.Column("id", sa.String(36), primary_key=True),
+        sa.Column("conversation_id", sa.String(36), nullable=False, index=True),
+        sa.Column("direction", sa.String(10), nullable=False),
+        sa.Column("role", sa.String(20), nullable=False),
+        sa.Column("message_type", sa.String(50), server_default="text"),
+        sa.Column("content", sa.Text(), nullable=False),
+        sa.Column("whatsapp_message_id", sa.String(100), nullable=True),
+        sa.Column("tool_calls", postgresql.JSONB(astext_type=sa.Text()), nullable=True),
+        sa.Column("tool_results", postgresql.JSONB(astext_type=sa.Text()), nullable=True),
+        sa.Column("tokens_used", sa.Integer(), server_default="0"),
+        sa.Column("metadata", postgresql.JSONB(astext_type=sa.Text()), server_default="{}"),
+        sa.Column("created_at", sa.DateTime(timezone=True), server_default=sa.func.now()),
+    )
+
+    # Knowledge chunks table (for RAG)
+    op.create_table(
+        "knowledge_chunks",
+        sa.Column("id", sa.String(36), primary_key=True, server_default=sa.text("gen_random_uuid()::text")),
+        sa.Column("content", sa.Text(), nullable=False),
+        sa.Column("metadata", postgresql.JSONB(astext_type=sa.Text()), server_default="{}"),
+        sa.Column("category", sa.String(50), server_default="general"),
+        sa.Column("source", sa.String(255), server_default=""),
+        sa.Column("embedding", sa.String(), nullable=True),  # Stored as string; pgvector uses special type
+        sa.Column("created_at", sa.DateTime(timezone=True), server_default=sa.func.now()),
+        sa.Column("updated_at", sa.DateTime(timezone=True), server_default=sa.func.now()),
+    )
+
+    # Create pgvector column properly using raw SQL
+    op.execute("""
+        ALTER TABLE knowledge_chunks
+        ALTER COLUMN embedding TYPE vector(1536)
+        USING embedding::vector(1536)
+    """)
+
+    # Indexes
+    op.create_index("idx_knowledge_category", "knowledge_chunks", ["category"])
+    op.create_index("idx_knowledge_source", "knowledge_chunks", ["source"])
+    op.execute("""
+        CREATE INDEX idx_knowledge_embedding
+        ON knowledge_chunks
+        USING ivfflat (embedding vector_cosine_ops)
+        WITH (lists = 100)
+    """)
+
+
+def downgrade() -> None:
+    op.drop_index("idx_knowledge_embedding", table_name="knowledge_chunks")
+    op.drop_index("idx_knowledge_source", table_name="knowledge_chunks")
+    op.drop_index("idx_knowledge_category", table_name="knowledge_chunks")
+    op.drop_table("knowledge_chunks")
+    op.drop_table("messages")
+    op.drop_table("conversations")
+    op.execute("DROP TYPE IF EXISTS conversationstatus")
+    op.execute("DROP EXTENSION IF EXISTS vector")
--- a/docker-compose.yml
+++ b/docker-compose.yml
@@ -0,0 +1,145 @@
+version: "3.9"
+
+services:
+  api:
+    build:
+      context: .
+      dockerfile: Dockerfile
+    container_name: skeen-api
+    restart: unless-stopped
+    ports:
+      - "8000:8000"
+    env_file:
+      - .env
+    environment:
+      - DATABASE_URL=postgresql+asyncpg://skeen:skeen123@postgres:5432/skeen_crm
+      - REDIS_URL=redis://redis:6379/0
+      - CELERY_BROKER_URL=redis://redis:6379/1
+      - CELERY_RESULT_BACKEND=redis://redis:6379/2
+    depends_on:
+      postgres:
+        condition: service_healthy
+      redis:
+        condition: service_started
+      migrate:
+        condition: service_completed_successfully
+    networks:
+      - skeen-network
+    command: >
+      uvicorn src.main:app
+      --host 0.0.0.0
+      --port 8000
+      --workers 2
+      --loop uvloop
+      --http httptools
+      --log-level info
+
+  worker:
+    build:
+      context: .
+      dockerfile: Dockerfile
+    container_name: skeen-worker
+    restart: unless-stopped
+    env_file:
+      - .env
+    environment:
+      - DATABASE_URL=postgresql+asyncpg://skeen:skeen123@postgres:5432/skeen_crm
+      - REDIS_URL=redis://redis:6379/0
+      - CELERY_BROKER_URL=redis://redis:6379/1
+      - CELERY_RESULT_BACKEND=redis://redis:6379/2
+    depends_on:
+      postgres:
+        condition: service_healthy
+      redis:
+        condition: service_started
+      migrate:
+        condition: service_completed_successfully
+    networks:
+      - skeen-network
+    command: >
+      celery -A src.workers.celery_app worker
+      --loglevel=info
+      --concurrency=4
+      -Q default,whatsapp,erpnext,ai
+
+  scheduler:
+    build:
+      context: .
+      dockerfile: Dockerfile
+    container_name: skeen-scheduler
+    restart: unless-stopped
+    env_file:
+      - .env
+    environment:
+      - DATABASE_URL=postgresql+asyncpg://skeen:skeen123@postgres:5432/skeen_crm
+      - REDIS_URL=redis://redis:6379/0
+      - CELERY_BROKER_URL=redis://redis:6379/1
+      - CELERY_RESULT_BACKEND=redis://redis:6379/2
+    depends_on:
+      postgres:
+        condition: service_healthy
+      redis:
+        condition: service_started
+    networks:
+      - skeen-network
+    command: >
+      celery -A src.workers.celery_app beat
+      --loglevel=info
+      --scheduler celery.beat.PersistentScheduler
+
+  migrate:
+    build:
+      context: .
+      dockerfile: Dockerfile
+    container_name: skeen-migrate
+    env_file:
+      - .env
+    environment:
+      - DATABASE_URL=postgresql+asyncpg://skeen:skeen123@postgres:5432/skeen_crm
+    depends_on:
+      postgres:
+        condition: service_healthy
+    networks:
+      - skeen-network
+    command: >
+      alembic upgrade head
+
+  postgres:
+    image: ankane/pgvector:latest
+    container_name: skeen-postgres
+    restart: unless-stopped
+    environment:
+      POSTGRES_USER: skeen
+      POSTGRES_PASSWORD: skeen123
+      POSTGRES_DB: skeen_crm
+    volumes:
+      - postgres_data:/var/lib/postgresql/data
+    ports:
+      - "5432:5432"
+    networks:
+      - skeen-network
+    healthcheck:
+      test: [ "CMD-SHELL", "pg_isready -U skeen -d skeen_crm" ]
+      interval: 5s
+      timeout: 5s
+      retries: 5
+
+  redis:
+    image: redis:7-alpine
+    container_name: skeen-redis
+    restart: unless-stopped
+    volumes:
+      - redis_data:/data
+    ports:
+      - "6379:6379"
+    networks:
+      - skeen-network
+    command: redis-server --appendonly yes
+
+volumes:
+  postgres_data:
+  redis_data:
+
+networks:
+  skeen-network:
+    driver: bridge
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -0,0 +1,102 @@
+[project]
+name = "skeen-crm-agent"
+version = "0.1.0"
+description = "SKEEN CRM AI Agent for WhatsApp Business API + ERPNext"
+readme = "README.md"
+requires-python = ">=3.12"
+dependencies = [
+    # Web Framework
+    "fastapi>=0.111.0",
+    "uvicorn[standard]>=0.30.0",
+    "python-multipart>=0.0.9",
+    "python-jose[cryptography]>=3.3.0",
+    "passlib[bcrypt]>=1.7.4",
+    "httpx[http2]>=0.27.0",
+    "tenacity>=8.3.0",
+
+    # Database
+    "sqlalchemy[asyncio]>=2.0.30",
+    "asyncpg>=0.29.0",
+    "sqlmodel>=0.0.19",
+    "alembic>=1.13.0",
+    "pgvector>=0.2.5",
+
+    # Redis & Celery
+    "redis>=5.0.0",
+    "celery>=5.4.0",
+
+    # AI / LLM
+    "openai>=1.30.0",
+    "tiktoken>=0.7.0",
+
+    # Configuration & Validation
+    "pydantic>=2.7.0",
+    "pydantic-settings>=2.2.0",
+    "email-validator>=2.1.0",
+
+    # Observability
+    "structlog>=24.1.0",
+    "prometheus-client>=0.20.0",
+    "asgi-correlation-id>=4.3.0",
+
+    # Utilities
+    "orjson>=3.10.0",
+    "python-dateutil>=2.9.0",
+    "pendulum>=3.0.0",
+]
+
+[project.optional-dependencies]
+dev = [
+    "pytest>=8.2.0",
+    "pytest-asyncio>=0.23.0",
+    "pytest-cov>=5.0.0",
+    "httpx>=0.27.0",
+    "respx>=0.21.0",
+    "ruff>=0.4.0",
+    "mypy>=1.10.0",
+    "pre-commit>=3.7.0",
+    "types-python-dateutil>=2.9.0",
+    "types-passlib>=1.7.0",
+]
+
+[build-system]
+requires = ["hatchling"]
+build-backend = "hatchling.build"
+
+[tool.ruff]
+target-version = "py312"
+line-length = 100
+select = [
+    "E", "F", "W", "I", "N", "D", "UP", "B", "C4", "SIM", "ARG",
+    "PTH", "ERA", "RUF", "ASYNC", "S", "C90",
+]
+ignore = ["D100", "D104", "D107", "S101"]
+
+[tool.ruff.pydocstyle]
+convention = "google"
+
+[tool.mypy]
+python_version = "3.12"
+strict = true
+warn_return_any = true
+warn_unused_configs = true
+ignore_missing_imports = true
+
+[tool.pytest.ini_options]
+asyncio_mode = "auto"
+testpaths = ["tests"]
+python_files = ["test_*.py"]
+addopts = "-v --tb=short --strict-markers"
+markers = [
+    "unit: Unit tests",
+    "integration: Integration tests",
+    "slow: Slow tests",
+]
+
+[tool.coverage.run]
+source = ["src"]
+omit = ["*/tests/*", "*/migrations/*"]
+
+[tool.coverage.report]
+precision = 2
+fail_under = 80
--- a/src/init.py
+++ b/src/init.py
--- a/src/api/init.py
+++ b/src/api/init.py
--- a/src/api/deps.py
+++ b/src/api/deps.py
@@ -0,0 +1,36 @@
+"""FastAPI dependency injection providers."""
+
+from typing import AsyncGenerator
+
+from fastapi import Request
+from sqlalchemy.ext.asyncio import AsyncSession
+
+from src.infrastructure.db import AsyncSessionLocal
+from src.infrastructure.redis import RedisCache, get_redis
+
+
+async def get_db_session() -> AsyncGenerator[AsyncSession, None]:
+    """Yield a database session for FastAPI dependency injection."""
+    session = AsyncSessionLocal()
+    try:
+        yield session
+        await session.commit()
+    except Exception:
+        await session.rollback()
+        raise
+    finally:
+        await session.close()
+
+
+async def get_cache() -> AsyncGenerator[RedisCache, None]:
+    """Yield a Redis cache instance."""
+    redis = await get_redis()
+    yield RedisCache(redis)
+
+
+def get_client_ip(request: Request) -> str:
+    """Extract client IP from request, handling proxies."""
+    forwarded = request.headers.get("x-forwarded-for")
+    if forwarded:
+        return forwarded.split(",")[0].strip()
+    return request.client.host if request.client else "unknown"
--- a/src/api/v1/init.py
+++ b/src/api/v1/init.py
--- a/src/api/v1/health.py
+++ b/src/api/v1/health.py
@@ -0,0 +1,39 @@
+"""Health check endpoints."""
+
+from datetime import datetime, timezone
+
+from fastapi import APIRouter, Depends, status
+from sqlalchemy import text
+from sqlalchemy.ext.asyncio import AsyncSession
+
+from src.api.deps import get_db_session
+from src.config import settings
+
+router = APIRouter(tags=["health"])
+
+
+@router.get("/health", status_code=status.HTTP_200_OK)
+async def health_check() -> dict:
+    """Liveness probe."""
+    return {
+        "status": "healthy",
+        "timestamp": datetime.now(timezone.utc).isoformat(),
+        "version": "0.1.0",
+        "environment": settings.APP_ENV,
+    }
+
+
+@router.get("/ready", status_code=status.HTTP_200_OK)
+async def readiness_check(db: AsyncSession = Depends(get_db_session)) -> dict:
+    """Readiness probe including database connectivity."""
+    try:
+        await db.execute(text("SELECT 1"))
+        db_status = "connected"
+    except Exception as exc:
+        db_status = f"error: {exc}"
+
+    return {
+        "status": "ready" if db_status == "connected" else "not_ready",
+        "database": db_status,
+        "timestamp": datetime.now(timezone.utc).isoformat(),
+    }
--- a/src/api/v1/messages.py
+++ b/src/api/v1/messages.py
@@ -0,0 +1,109 @@
+"""API endpoints for sending WhatsApp messages manually."""
+
+from fastapi import APIRouter, Depends, HTTPException, status
+from pydantic import BaseModel, Field
+from sqlalchemy.ext.asyncio import AsyncSession
+
+from src.api.deps import get_db_session
+from src.infrastructure.whatsapp.client import get_whatsapp_client
+
+router = APIRouter(prefix="/messages", tags=["messages"])
+
+
+class SendTextMessageRequest(BaseModel):
+    to: str = Field(..., description="Phone number in international format (e.g., 5216641234567)")
+    text: str = Field(..., max_length=4096, description="Message text")
+    preview_url: bool = Field(False, description="Show URL preview")
+
+
+class SendTemplateMessageRequest(BaseModel):
+    to: str = Field(..., description="Phone number in international format")
+    template_name: str = Field(..., description="Registered template name")
+    language_code: str = Field("es_MX", description="Template language code")
+
+
+class SendButtonsRequest(BaseModel):
+    to: str
+    body: str = Field(..., max_length=1024)
+    buttons: list[dict[str, str]] = Field(..., min_length=1, max_length=3)
+
+
+class MessageResponse(BaseModel):
+    status: str
+    message_id: str | None = None
+    details: dict | None = None
+
+
+@router.post("/text", response_model=MessageResponse, status_code=status.HTTP_201_CREATED)
+async def send_text_message(
+    request: SendTextMessageRequest,
+    db: AsyncSession = Depends(get_db_session),
+) -> MessageResponse:
+    """Send a text message to a WhatsApp user."""
+    client = await get_whatsapp_client()
+    try:
+        result = await client.send_text_message(
+            to=request.to,
+            text=request.text,
+            preview_url=request.preview_url,
+        )
+        return MessageResponse(
+            status="sent",
+            message_id=result.get("messages", [{}])[0].get("id"),
+            details=result,
+        )
+    except Exception as exc:
+        raise HTTPException(
+            status_code=status.HTTP_502_BAD_GATEWAY,
+            detail=str(exc),
+        ) from exc
+
+
+@router.post("/template", response_model=MessageResponse, status_code=status.HTTP_201_CREATED)
+async def send_template_message(
+    request: SendTemplateMessageRequest,
+    db: AsyncSession = Depends(get_db_session),
+) -> MessageResponse:
+    """Send a template message (for 24h+ window)."""
+    client = await get_whatsapp_client()
+    try:
+        result = await client.send_template_message(
+            to=request.to,
+            template_name=request.template_name,
+            language_code=request.language_code,
+        )
+        return MessageResponse(
+            status="sent",
+            message_id=result.get("messages", [{}])[0].get("id"),
+            details=result,
+        )
+    except Exception as exc:
+        raise HTTPException(
+            status_code=status.HTTP_502_BAD_GATEWAY,
+            detail=str(exc),
+        ) from exc
+
+
+@router.post("/buttons", response_model=MessageResponse, status_code=status.HTTP_201_CREATED)
+async def send_buttons(
+    request: SendButtonsRequest,
+    db: AsyncSession = Depends(get_db_session),
+) -> MessageResponse:
+    """Send an interactive button message."""
+    client = await get_whatsapp_client()
+    try:
+        result = await client.send_interactive_buttons(
+            to=request.to,
+            body=request.body,
+            buttons=request.buttons,
+        )
+        return MessageResponse(
+            status="sent",
+            message_id=result.get("messages", [{}])[0].get("id"),
+            details=result,
+        )
+    except Exception as exc:
+        raise HTTPException(
+            status_code=status.HTTP_502_BAD_GATEWAY,
+            detail=str(exc),
+        ) from exc
--- a/src/api/v1/webhooks.py
+++ b/src/api/v1/webhooks.py
@@ -0,0 +1,98 @@
+"""Meta WhatsApp webhook endpoints."""
+
+from fastapi import APIRouter, Depends, Header, HTTPException, Request, status
+from pydantic import BaseModel
+from sqlalchemy.ext.asyncio import AsyncSession
+
+from src.api.deps import get_client_ip, get_db_session
+from src.config import settings
+from src.core.exceptions import ValidationError
+from src.infrastructure.whatsapp.webhook import (
+    WebhookVerifier,
+    parse_webhook_payload,
+)
+from src.use_cases.handle_incoming_message import process_incoming_message
+from src.workers.celery_app import process_whatsapp_message_task
+
+router = APIRouter(prefix="/webhooks", tags=["webhooks"])
+
+
+class WebhookVerificationResponse(BaseModel):
+    challenge: str
+
+
+@router.get("/whatsapp")
+async def verify_whatsapp_webhook(
+    hub_mode: str | None = None,
+    hub_verify_token: str | None = None,
+    hub_challenge: str | None = None,
+) -> str:
+    """Verify webhook subscription with Meta.
+
+    Meta sends a GET request to verify the endpoint during webhook setup.
+    """
+    is_valid = WebhookVerifier.verify_subscription(hub_mode, hub_verify_token)
+    if not is_valid:
+        raise HTTPException(
+            status_code=status.HTTP_403_FORBIDDEN,
+            detail="Invalid verification token",
+        )
+    return hub_challenge or ""
+
+
+@router.post("/whatsapp", status_code=status.HTTP_200_OK)
+async def receive_whatsapp_webhook(
+    request: Request,
+    db: AsyncSession = Depends(get_db_session),
+    x_hub_signature_256: str | None = Header(None),
+    x_forwarded_for: str | None = Header(None),
+) -> dict:
+    """Receive incoming WhatsApp messages and status updates.
+
+    In production, this endpoint should return 200 immediately and
+    process the message asynchronously via Celery to avoid timeouts.
+    """
+    body = await request.body()
+
+    # Verify signature in production
+    if settings.is_production:
+        is_valid = WebhookVerifier.verify_signature(body, x_hub_signature_256)
+        if not is_valid:
+            raise HTTPException(
+                status_code=status.HTTP_401_UNAUTHORIZED,
+                detail="Invalid signature",
+            )
+
+    try:
+        payload = await request.json()
+        webhook = parse_webhook_payload(payload)
+    except ValidationError as exc:
+        # Return 200 for non-message events to prevent Meta retries
+        return {"status": "ignored", "reason": exc.message}
+    except Exception:
+        return {"status": "ignored", "reason": "invalid_payload"}
+
+    client_ip = x_forwarded_for or get_client_ip(request)
+
+    # Handle message statuses (delivered, read, etc.)
+    if webhook.has_statuses:
+        # TODO: Update message status in database
+        return {"status": "acknowledged", "type": "status_update"}
+
+    # Handle incoming messages
+    if webhook.has_messages:
+        if settings.is_production:
+            # Async processing via Celery for production
+            process_whatsapp_message_task.delay(
+                webhook.raw,
+                client_ip=client_ip,
+            )
+        else:
+            # Sync processing for development/testing
+            await process_incoming_message(
+                db=db,
+                webhook_data=webhook.raw,
+                client_ip=client_ip,
+            )
+
+    return {"status": "processed"}
--- a/src/config.py
+++ b/src/config.py
@@ -0,0 +1,96 @@
+"""Application configuration using Pydantic Settings."""
+
+from functools import lru_cache
+
+from pydantic import Field, PostgresDsn, RedisDsn, SecretStr
+from pydantic_settings import BaseSettings, SettingsConfigDict
+
+
+class Settings(BaseSettings):
+    """Application settings loaded from environment variables."""
+
+    model_config = SettingsConfigDict(
+        env_file=".env",
+        env_file_encoding="utf-8",
+        extra="ignore",
+    )
+
+    # App
+    APP_NAME: str = "Skeen-CRM-Agent"
+    APP_ENV: str = "development"
+    DEBUG: bool = False
+    LOG_LEVEL: str = "INFO"
+    SECRET_KEY: SecretStr = SecretStr("change-me")
+
+    # Server
+    HOST: str = "0.0.0.0"
+    PORT: int = 8000
+
+    # Database
+    DATABASE_URL: PostgresDsn = PostgresDsn(
+        "postgresql+asyncpg://skeen:skeen123@localhost:5432/skeen_crm"
+    )
+    DATABASE_POOL_SIZE: int = 20
+    DATABASE_MAX_OVERFLOW: int = 10
+
+    # Redis
+    REDIS_URL: RedisDsn = RedisDsn("redis://localhost:6379/0")
+
+    # Meta / WhatsApp Business API
+    META_API_VERSION: str = "v18.0"
+    META_ACCESS_TOKEN: SecretStr = SecretStr("")
+    META_PHONE_NUMBER_ID: str = ""
+    META_BUSINESS_ACCOUNT_ID: str = ""
+    META_WEBHOOK_VERIFY_TOKEN: SecretStr = SecretStr("")
+    META_APP_SECRET: SecretStr = SecretStr("")
+
+    # OpenAI
+    OPENAI_API_KEY: SecretStr = SecretStr("")
+    OPENAI_MODEL: str = "gpt-4o"
+    OPENAI_EMBEDDING_MODEL: str = "text-embedding-3-small"
+    OPENAI_TEMPERATURE: float = 0.3
+    OPENAI_MAX_TOKENS: int = 1500
+
+    # RAG
+    VECTOR_DIMENSION: int = 1536
+    RAG_TOP_K: int = 5
+    RAG_SIMILARITY_THRESHOLD: float = 0.75
+
+    # ERPNext
+    ERPNEXT_BASE_URL: str = ""
+    ERPNEXT_API_KEY: SecretStr = SecretStr("")
+    ERPNEXT_API_SECRET: SecretStr = SecretStr("")
+    ERPNEXT_VERIFY_SSL: bool = True
+
+    # Celery
+    CELERY_BROKER_URL: RedisDsn = RedisDsn("redis://localhost:6379/1")
+    CELERY_RESULT_BACKEND: RedisDsn = RedisDsn("redis://localhost:6379/2")
+    CELERY_WORKER_CONCURRENCY: int = 4
+
+    # Monitoring
+    ENABLE_METRICS: bool = True
+    SENTRY_DSN: str = ""
+
+    @property
+    def database_url_str(self) -> str:
+        """Return database URL as plain string for SQLAlchemy."""
+        return str(self.DATABASE_URL)
+
+    @property
+    def meta_api_base_url(self) -> str:
+        """Return Meta Graph API base URL."""
+        return f"https://graph.facebook.com/{self.META_API_VERSION}"
+
+    @property
+    def is_production(self) -> bool:
+        """Check if running in production environment."""
+        return self.APP_ENV.lower() == "production"
+
+
+@lru_cache
+def get_settings() -> Settings:
+    """Return cached settings instance."""
+    return Settings()
+
+
+settings = get_settings()
--- a/src/core/init.py
+++ b/src/core/init.py
--- a/src/core/constants.py
+++ b/src/core/constants.py
@@ -0,0 +1,100 @@
+"""Application-wide constants."""
+
+from enum import Enum
+
+
+class MessageRole(str, Enum):
+    """Conversation message roles."""
+
+    SYSTEM = "system"
+    USER = "user"
+    ASSISTANT = "assistant"
+    TOOL = "tool"
+
+
+class ConversationStatus(str, Enum):
+    """Status of a patient conversation."""
+
+    ACTIVE = "active"
+    PAUSED = "paused"
+    RESOLVED = "resolved"
+    ESCALATED = "escalated"
+    APPOINTMENT_CONFIRMED = "appointment_confirmed"
+
+
+class AppointmentType(str, Enum):
+    """Types of medical appointments."""
+
+    PRIMERA_VEZ = "primera_vez"
+    SUBSECUENTE = "subsecuente"
+    PAQUETE = "paquete"
+    VALORACION = "valoracion"
+
+
+class PaymentMethod(str, Enum):
+    """Supported payment methods."""
+
+    EFECTIVO_MN = "efectivo_mn"
+    EFECTIVO_USD = "efectivo_usd"
+    TARJETA = "tarjeta"
+    TRANSFERENCIA = "transferencia"
+    MONEDERO = "monedero"
+
+
+class WhatsAppMessageType(str, Enum):
+    """WhatsApp message types from Meta webhook."""
+
+    TEXT = "text"
+    IMAGE = "image"
+    AUDIO = "audio"
+    VIDEO = "video"
+    DOCUMENT = "document"
+    LOCATION = "location"
+    CONTACTS = "contacts"
+    INTERACTIVE = "interactive"
+    BUTTON_REPLY = "button_reply"
+    LIST_REPLY = "list_reply"
+    UNKNOWN = "unknown"
+
+
+class WhatsAppMessageStatus(str, Enum):
+    """WhatsApp message delivery statuses."""
+
+    SENT = "sent"
+    DELIVERED = "delivered"
+    READ = "read"
+    FAILED = "failed"
+
+
+# System prompts
+SKEEN_SYSTEM_PROMPT = """Eres SKEEN Assistant, el agente de inteligencia artificial oficial de SKEEN Clínica de Belleza y Dermatología Estética.
+
+INFORMACIÓN DEL NEGOCIO:
+- SKEEN es una clínica dermatológica y estética con sedes en Rosarito y Tijuana, B.C., México.
+- Ofrecemos tratamientos faciales, corporales, depilación láser, toxina botulínica, ácido hialurónico, y consultas dermatológicas.
+- Horario de atención: Lunes a Sábado 9:00 - 18:00, Domingos 10:00 - 14:00.
+- Teléfono: (664) 123-4567
+
+TU ROL:
+1. Atiende a pacientes y prospectos por WhatsApp de manera cálida, profesional y eficiente.
+2. Agenda citas verificando disponibilidad con el CRM.
+3. Responde preguntas sobre servicios, precios, paquetes y promociones usando el catálogo (RAG).
+4. Procesa pagos y consulta saldo de monedero electrónico.
+5. Escal a un humano cuando el paciente lo solicite o el caso sea complejo.
+
+REGLAS CRÍTICAS:
+- SIEMPRE saluda por el nombre del paciente si lo conoces.
+- NUNCA inventes precios ni disponibilidad. Consulta el RAG o el CRM.
+- SIEMPRE confirma los detalles de la cita (fecha, hora, sucursal, servicio, doctor) antes de agendar.
+- NUNCA compartas información médica de otros pacientes.
+- Usa emojis con moderación para mantener un tono cálido pero profesional.
+- El idioma principal es español (México).
+
+FORMATO DE RESPUESTA:
+- Sé conciso pero completo (máximo 3 párrafos cortos para WhatsApp).
+- Usa viñetas cuando listes opciones.
+- Incluye llamados a la acción claros.
+"""
+
+# Webhook
+WHATSAPP_WEBHOOK_SUBSCRIBE_MODE = "subscribe"
--- a/src/core/exceptions.py
+++ b/src/core/exceptions.py
@@ -0,0 +1,66 @@
+"""Custom application exceptions."""
+
+
+class SkeenException(Exception):
+    """Base exception for SKEEN application."""
+
+    def __init__(self, message: str, status_code: int = 500, details: dict | None = None) -> None:
+        self.message = message
+        self.status_code = status_code
+        self.details = details or {}
+        super().__init__(self.message)
+
+
+class WhatsAppAPIError(SkeenException):
+    """Error communicating with Meta WhatsApp Business API."""
+
+    def __init__(self, message: str, status_code: int = 502, details: dict | None = None) -> None:
+        super().__init__(message, status_code, details)
+
+
+class OpenAIError(SkeenException):
+    """Error communicating with OpenAI API."""
+
+    def __init__(self, message: str, status_code: int = 502, details: dict | None = None) -> None:
+        super().__init__(message, status_code, details)
+
+
+class ERPNextError(SkeenException):
+    """Error communicating with ERPNext API."""
+
+    def __init__(self, message: str, status_code: int = 502, details: dict | None = None) -> None:
+        super().__init__(message, status_code, details)
+
+
+class ConversationNotFoundError(SkeenException):
+    """Requested conversation does not exist."""
+
+    def __init__(self, conversation_id: str) -> None:
+        super().__init__(
+            f"Conversation {conversation_id} not found",
+            status_code=404,
+        )
+
+
+class PatientNotFoundError(SkeenException):
+    """Requested patient does not exist."""
+
+    def __init__(self, patient_id: str) -> None:
+        super().__init__(
+            f"Patient {patient_id} not found",
+            status_code=404,
+        )
+
+
+class ValidationError(SkeenException):
+    """Input validation error."""
+
+    def __init__(self, message: str, details: dict | None = None) -> None:
+        super().__init__(message, status_code=422, details=details)
+
+
+class RateLimitError(SkeenException):
+    """Rate limit exceeded."""
+
+    def __init__(self, message: str = "Rate limit exceeded") -> None:
+        super().__init__(message, status_code=429)
--- a/src/domain/init.py
+++ b/src/domain/init.py
--- a/src/domain/models/init.py
+++ b/src/domain/models/init.py
--- a/src/domain/models/conversation.py
+++ b/src/domain/models/conversation.py
@@ -0,0 +1,74 @@
+"""Conversation and message domain models."""
+
+from datetime import datetime, timezone
+from typing import Any
+
+from sqlalchemy import JSON, Column, DateTime, Enum, String, Text
+from sqlalchemy.dialects.postgresql import UUID
+from sqlalchemy.orm import Mapped, mapped_column
+from sqlalchemy.sql import func
+
+from src.core.constants import ConversationStatus
+from src.infrastructure.db import Base
+
+
+class Conversation(Base):
+    """A WhatsApp conversation with a patient."""
+
+    __tablename__ = "conversations"
+
+    id: Mapped[str] = mapped_column(String(36), primary_key=True)
+    phone_number: Mapped[str] = mapped_column(String(20), index=True, nullable=False)
+    patient_id: Mapped[str | None] = mapped_column(String(100), index=True, nullable=True)
+    patient_name: Mapped[str | None] = mapped_column(String(255), nullable=True)
+    status: Mapped[ConversationStatus] = mapped_column(
+        Enum(ConversationStatus),
+        default=ConversationStatus.ACTIVE,
+        nullable=False,
+    )
+    context: Mapped[dict[str, Any]] = mapped_column(JSON, default=dict)
+    last_message_at: Mapped[datetime | None] = mapped_column(
+        DateTime(timezone=True),
+        nullable=True,
+    )
+    created_at: Mapped[datetime] = mapped_column(
+        DateTime(timezone=True),
+        server_default=func.now(),
+    )
+    updated_at: Mapped[datetime] = mapped_column(
+        DateTime(timezone=True),
+        server_default=func.now(),
+        onupdate=func.now(),
+    )
+
+
+class Message(Base):
+    """Individual message within a conversation."""
+
+    __tablename__ = "messages"
+
+    id: Mapped[str] = mapped_column(String(36), primary_key=True)
+    conversation_id: Mapped[str] = mapped_column(
+        String(36),
+        index=True,
+        nullable=False,
+    )
+    direction: Mapped[str] = mapped_column(
+        String(10),
+        nullable=False,
+    )  # 'inbound' or 'outbound'
+    role: Mapped[str] = mapped_column(String(20), nullable=False)
+    message_type: Mapped[str] = mapped_column(String(50), default="text")
+    content: Mapped[str] = mapped_column(Text, nullable=False)
+    whatsapp_message_id: Mapped[str | None] = mapped_column(
+        String(100),
+        nullable=True,
+    )
+    tool_calls: Mapped[list[dict[str, Any]] | None] = mapped_column(JSON, nullable=True)
+    tool_results: Mapped[list[dict[str, Any]] | None] = mapped_column(JSON, nullable=True)
+    tokens_used: Mapped[int | None] = mapped_column(default=0)
+    metadata: Mapped[dict[str, Any]] = mapped_column(JSON, default=dict)
+    created_at: Mapped[datetime] = mapped_column(
+        DateTime(timezone=True),
+        server_default=func.now(),
+    )
--- a/src/domain/services/init.py
+++ b/src/domain/services/init.py
--- a/src/infrastructure/init.py
+++ b/src/infrastructure/init.py
--- a/src/infrastructure/ai/init.py
+++ b/src/infrastructure/ai/init.py
--- a/src/infrastructure/ai/openai_client.py
+++ b/src/infrastructure/ai/openai_client.py
@@ -0,0 +1,143 @@
+"""Async OpenAI client with structured logging and retry logic."""
+
+import json
+from typing import Any
+
+import openai
+import structlog
+from openai import AsyncOpenAI
+from tenacity import retry, retry_if_exception_type, stop_after_attempt, wait_exponential
+
+from src.config import settings
+from src.core.exceptions import OpenAIError
+
+logger = structlog.get_logger(__name__)
+
+
+class OpenAIClient:
+    """Enterprise-grade async OpenAI client."""
+
+    def __init__(self) -> None:
+        self.client = AsyncOpenAI(
+            api_key=settings.OPENAI_API_KEY.get_secret_value(),
+            max_retries=0,  # We handle retries manually with tenacity
+        )
+        self.model = settings.OPENAI_MODEL
+        self.embedding_model = settings.OPENAI_EMBEDDING_MODEL
+        self.temperature = settings.OPENAI_TEMPERATURE
+        self.max_tokens = settings.OPENAI_MAX_TOKENS
+
+    @retry(
+        stop=stop_after_attempt(3),
+        wait=wait_exponential(multiplier=1, min=2, max=10),
+        retry=retry_if_exception_type((openai.RateLimitError, openai.APITimeoutError)),
+        reraise=True,
+    )
+    async def chat_completion(
+        self,
+        messages: list[dict[str, str]],
+        tools: list[dict[str, Any]] | None = None,
+        tool_choice: str | dict[str, Any] = "auto",
+        temperature: float | None = None,
+        max_tokens: int | None = None,
+    ) -> dict[str, Any]:
+        """Create a chat completion with optional function calling."""
+        try:
+            response = await self.client.chat.completions.create(
+                model=self.model,
+                messages=messages,  # type: ignore[arg-type]
+                tools=tools,  # type: ignore[arg-type]
+                tool_choice=tool_choice if tools else None,  # type: ignore[arg-type]
+                temperature=temperature or self.temperature,
+                max_tokens=max_tokens or self.max_tokens,
+            )
+
+            message = response.choices[0].message
+            result = {
+                "content": message.content,
+                "role": message.role,
+                "tool_calls": None,
+                "finish_reason": response.choices[0].finish_reason,
+                "usage": {
+                    "prompt_tokens": response.usage.prompt_tokens if response.usage else 0,
+                    "completion_tokens": response.usage.completion_tokens if response.usage else 0,
+                    "total_tokens": response.usage.total_tokens if response.usage else 0,
+                },
+            }
+
+            if message.tool_calls:
+                result["tool_calls"] = [
+                    {
+                        "id": tc.id,
+                        "type": tc.type,
+                        "function": {
+                            "name": tc.function.name,
+                            "arguments": tc.function.arguments,
+                        },
+                    }
+                    for tc in message.tool_calls
+                ]
+
+            logger.info(
+                "openai_chat_completion",
+                model=self.model,
+                finish_reason=result["finish_reason"],
+                tokens=result["usage"]["total_tokens"],
+            )
+            return result
+
+        except openai.RateLimitError as exc:
+            logger.error("openai_rate_limited", error=str(exc))
+            raise OpenAIError("Rate limited by OpenAI", status_code=429) from exc
+        except openai.AuthenticationError as exc:
+            logger.error("openai_auth_error", error=str(exc))
+            raise OpenAIError("OpenAI authentication failed", status_code=401) from exc
+        except openai.APIError as exc:
+            logger.error("openai_api_error", error=str(exc), code=exc.code)
+            raise OpenAIError(f"OpenAI API error: {exc}", status_code=502) from exc
+
+    @retry(
+        stop=stop_after_attempt(3),
+        wait=wait_exponential(multiplier=1, min=1, max=5),
+        retry=retry_if_exception_type((openai.RateLimitError,)),
+        reraise=True,
+    )
+    async def create_embedding(self, text: str) -> list[float]:
+        """Create embedding vector for text."""
+        try:
+            response = await self.client.embeddings.create(
+                model=self.embedding_model,
+                input=text,
+                encoding_format="float",
+            )
+            embedding = response.data[0].embedding
+            logger.info(
+                "openai_embedding_created",
+                model=self.embedding_model,
+                dimensions=len(embedding),
+            )
+            return embedding
+        except openai.APIError as exc:
+            logger.error("openai_embedding_error", error=str(exc))
+            raise OpenAIError(f"Embedding failed: {exc}", status_code=502) from exc
+
+    async def create_embeddings_batch(self, texts: list[str]) -> list[list[float]]:
+        """Create embeddings for multiple texts."""
+        response = await self.client.embeddings.create(
+            model=self.embedding_model,
+            input=texts,
+            encoding_format="float",
+        )
+        return [d.embedding for d in response.data]
+
+
+# Global singleton
+_openai_client: OpenAIClient | None = None
+
+
+async def get_openai_client() -> OpenAIClient:
+    """Return OpenAI client singleton."""
+    global _openai_client
+    if _openai_client is None:
+        _openai_client = OpenAIClient()
+    return _openai_client
--- a/src/infrastructure/ai/prompts.py
+++ b/src/infrastructure/ai/prompts.py
@@ -0,0 +1,202 @@
+"""Prompt templates and tool definitions for the SKEEN AI Agent."""
+
+from src.core.constants import SKEEN_SYSTEM_PROMPT
+
+# =============================================================================
+# TOOL DEFINITIONS (OpenAI Function Calling)
+# =============================================================================
+
+TOOLS = [
+    {
+        "type": "function",
+        "function": {
+            "name": "search_catalog",
+            "description": "Busca servicios, productos o paquetes en el catálogo de SKEEN. Usa esta herramienta cuando el paciente pregunte por tratamientos, precios, disponibilidad o promociones.",
+            "parameters": {
+                "type": "object",
+                "properties": {
+                    "query": {
+                        "type": "string",
+                        "description": "Términos de búsqueda del paciente (ej: 'depilación láser bikini', 'toxina botulínica precio')",
+                    },
+                    "category": {
+                        "type": "string",
+                        "enum": ["servicio", "producto", "paquete", "general"],
+                        "description": "Categoría a filtrar, si se menciona explícitamente",
+                    },
+                },
+                "required": ["query"],
+            },
+        },
+    },
+    {
+        "type": "function",
+        "function": {
+            "name": "check_availability",
+            "description": "Consulta la disponibilidad de citas en una fecha, sucursal o con un doctor específico.",
+            "parameters": {
+                "type": "object",
+                "properties": {
+                    "date": {
+                        "type": "string",
+                        "description": "Fecha en formato ISO 8601 (YYYY-MM-DD). Si no se especifica, usa la fecha de mañana.",
+                    },
+                    "branch": {
+                        "type": "string",
+                        "description": "Nombre de la sucursal: 'Rosarito' o 'Tijuana'. Si no se especifica, busca en ambas.",
+                    },
+                    "doctor": {
+                        "type": "string",
+                        "description": "Nombre del doctor o 'cualquiera'.",
+                    },
+                    "service": {
+                        "type": "string",
+                        "description": "Nombre del servicio que desea agendar.",
+                    },
+                },
+                "required": ["date"],
+            },
+        },
+    },
+    {
+        "type": "function",
+        "function": {
+            "name": "create_appointment",
+            "description": "Crea una cita médica en el sistema. SOLO usar después de confirmar fecha, hora, sucursal, servicio y doctor con el paciente.",
+            "parameters": {
+                "type": "object",
+                "properties": {
+                    "patient_phone": {
+                        "type": "string",
+                        "description": "Número de WhatsApp del paciente (con lada, ej: 5216641234567).",
+                    },
+                    "patient_name": {
+                        "type": "string",
+                        "description": "Nombre completo del paciente.",
+                    },
+                    "date": {
+                        "type": "string",
+                        "description": "Fecha de la cita (YYYY-MM-DD).",
+                    },
+                    "time": {
+                        "type": "string",
+                        "description": "Hora de la cita en formato 24h (HH:MM).",
+                    },
+                    "service": {
+                        "type": "string",
+                        "description": "Nombre del servicio a agendar.",
+                    },
+                    "branch": {
+                        "type": "string",
+                        "description": "Sucursal: 'Rosarito' o 'Tijuana'.",
+                    },
+                    "doctor": {
+                        "type": "string",
+                        "description": "Nombre del doctor asignado.",
+                    },
+                    "notes": {
+                        "type": "string",
+                        "description": "Notas adicionales para la cita.",
+                    },
+                },
+                "required": ["patient_phone", "patient_name", "date", "time", "service", "branch"],
+            },
+        },
+    },
+    {
+        "type": "function",
+        "function": {
+            "name": "get_patient_info",
+            "description": "Consulta la información de un paciente existente: historial de citas, saldo de monedero, adeudos, paquetes activos.",
+            "parameters": {
+                "type": "object",
+                "properties": {
+                    "phone": {
+                        "type": "string",
+                        "description": "Número de WhatsApp del paciente.",
+                    },
+                },
+                "required": ["phone"],
+            },
+        },
+    },
+    {
+        "type": "function",
+        "function": {
+            "name": "get_wallet_balance",
+            "description": "Consulta el saldo del monedero electrónico de un paciente.",
+            "parameters": {
+                "type": "object",
+                "properties": {
+                    "phone": {
+                        "type": "string",
+                        "description": "Número de WhatsApp del paciente.",
+                    },
+                },
+                "required": ["phone"],
+            },
+        },
+    },
+    {
+        "type": "function",
+        "function": {
+            "name": "escalate_to_human",
+            "description": "Escalar la conversación a un agente humano cuando el paciente lo solicite o el caso sea muy complejo/emocional.",
+            "parameters": {
+                "type": "object",
+                "properties": {
+                    "reason": {
+                        "type": "string",
+                        "description": "Motivo de la escalación.",
+                    },
+                },
+                "required": ["reason"],
+            },
+        },
+    },
+]
+
+# =============================================================================
+# PROMPT TEMPLATES
+# =============================================================================
+
+APPOINTMENT_CONFIRMATION_PROMPT = """El paciente ha confirmado los siguientes datos para su cita:
+
+- Nombre: {patient_name}
+- Fecha: {date}
+- Hora: {time}
+- Servicio: {service}
+- Sucursal: {branch}
+- Doctor: {doctor}
+
+Genera un mensaje de confirmación cálido y profesional para WhatsApp que:
+1. Agradezca la preferencia.
+2. Confirme todos los detalles en formato claro.
+3. Indique política de cancelación (24h de anticipación).
+4. Incluya dirección de la sucursal.
+5. Sea corto (máximo 4 párrafos)."""
+
+NO_AVAILABILITY_PROMPT = """No hay disponibilidad para la fecha/hora/sucursal solicitada.
+
+Sugerencias actuales:
+{alternatives}
+
+Genera un mensaje amable ofreciendo las alternativas disponibles. Mantén tono profesional y empático."""
+
+WALLET_BALANCE_PROMPT = """Saldo de monedero del paciente:
+- Saldo actual: ${balance_mxn} MXN
+- Puntos acumulados: {points}
+
+Genera un resumen amigable del saldo y sugiere cómo puede usarlo (aplicar a su próxima cita o paquete)."""
+
+NEW_PATIENT_ONBOARDING_PROMPT = """Este es un nuevo prospecto. Bienvenida/da a SKEEN.
+
+Información capturada:
+- Nombre: {name}
+- Interés: {interest}
+
+Genera un mensaje de bienvenida cálido que:
+1. Presente brevemente a SKEEN.
+2. Mencione que pueden agendar valoración gratuita.
+3. Pregunte si desea información sobre algún tratamiento específico.
+4. Incluya link a valoración express si es relevante."""
--- a/src/infrastructure/ai/rag.py
+++ b/src/infrastructure/ai/rag.py
@@ -0,0 +1,181 @@
+"""Retrieval-Augmented Generation with pgvector."""
+
+from typing import Any
+
+import structlog
+from sqlalchemy import text
+from sqlalchemy.ext.asyncio import AsyncSession
+
+from src.config import settings
+from src.infrastructure.ai.openai_client import get_openai_client
+
+logger = structlog.get_logger(__name__)
+
+
+class RAGStore:
+    """Vector store for SKEEN catalog and knowledge base using pgvector."""
+
+    def __init__(self, session: AsyncSession) -> None:
+        self.session = session
+
+    async def ensure_extension(self) -> None:
+        """Ensure pgvector extension is installed."""
+        await self.session.execute(text("CREATE EXTENSION IF NOT EXISTS vector"))
+        await self.session.commit()
+
+    async def search(
+        self,
+        query: str,
+        top_k: int | None = None,
+        similarity_threshold: float | None = None,
+        category: str | None = None,
+    ) -> list[dict[str, Any]]:
+        """Search knowledge base using semantic similarity.
+
+        Args:
+            query: User query text.
+            top_k: Number of results (default from settings).
+            similarity_threshold: Minimum similarity score.
+            category: Filter by category (e.g., 'servicio', 'producto', 'faq').
+
+        Returns:
+            List of matching chunks with content, metadata, and similarity score.
+        """
+        top_k = top_k or settings.RAG_TOP_K
+        threshold = similarity_threshold or settings.RAG_SIMILARITY_THRESHOLD
+
+        # Generate embedding for query
+        openai_client = await get_openai_client()
+        embedding = await openai_client.create_embedding(query)
+        embedding_str = f"[{','.join(str(x) for x in embedding)}]"
+
+        # Build query
+        category_filter = "AND category = :category" if category else ""
+        sql = f"""
+            SELECT
+                id,
+                content,
+                metadata,
+                category,
+                source,
+                1 - (embedding <=> :embedding_vec::vector) AS similarity
+            FROM knowledge_chunks
+            WHERE 1 - (embedding <=> :embedding_vec::vector) >= :threshold
+            {category_filter}
+            ORDER BY embedding <=> :embedding_vec::vector
+            LIMIT :top_k
+        """
+
+        params = {
+            "embedding_vec": embedding_str,
+            "threshold": threshold,
+            "top_k": top_k,
+        }
+        if category:
+            params["category"] = category
+
+        result = await self.session.execute(text(sql), params)
+        rows = result.mappings().all()
+
+        documents = []
+        for row in rows:
+            documents.append({
+                "id": row["id"],
+                "content": row["content"],
+                "metadata": row["metadata"],
+                "category": row["category"],
+                "source": row["source"],
+                "similarity": float(row["similarity"]),
+            })
+
+        logger.info(
+            "rag_search_completed",
+            query=query[:50],
+            results=len(documents),
+            category=category,
+        )
+        return documents
+
+    async def add_document(
+        self,
+        content: str,
+        metadata: dict[str, Any] | None = None,
+        category: str = "general",
+        source: str = "",
+        doc_id: str | None = None,
+    ) -> str:
+        """Add a document chunk to the knowledge base."""
+        openai_client = await get_openai_client()
+        embedding = await openai_client.create_embedding(content)
+        embedding_str = f"[{','.join(str(x) for x in embedding)}]"
+
+        sql = """
+            INSERT INTO knowledge_chunks (id, content, metadata, category, source, embedding)
+            VALUES (
+                COALESCE(:doc_id, gen_random_uuid()::text),
+                :content,
+                :metadata,
+                :category,
+                :source,
+                :embedding_vec::vector
+            )
+            RETURNING id
+        """
+
+        result = await self.session.execute(
+            text(sql),
+            {
+                "doc_id": doc_id,
+                "content": content,
+                "metadata": json.dumps(metadata or {}),
+                "category": category,
+                "source": source,
+                "embedding_vec": embedding_str,
+            },
+        )
+        await self.session.commit()
+        row = result.mappings().first()
+        inserted_id = row["id"] if row else ""
+
+        logger.info(
+            "rag_document_added",
+            doc_id=inserted_id,
+            category=category,
+            source=source,
+        )
+        return inserted_id
+
+    async def delete_by_source(self, source: str) -> int:
+        """Delete all chunks from a specific source."""
+        result = await self.session.execute(
+            text("DELETE FROM knowledge_chunks WHERE source = :source"),
+            {"source": source},
+        )
+        await self.session.commit()
+        deleted = result.rowcount or 0
+        logger.info("rag_documents_deleted", source=source, count=deleted)
+        return deleted
+
+
+# SQL to create table (run via Alembic or init script)
+CREATE_KNOWLEDGE_TABLE_SQL = """
+CREATE EXTENSION IF NOT EXISTS vector;
+
+CREATE TABLE IF NOT EXISTS knowledge_chunks (
+    id TEXT PRIMARY KEY DEFAULT gen_random_uuid()::text,
+    content TEXT NOT NULL,
+    metadata JSONB DEFAULT '{}',
+    category TEXT DEFAULT 'general',
+    source TEXT DEFAULT '',
+    embedding VECTOR(1536),
+    created_at TIMESTAMP WITH TIME ZONE DEFAULT NOW(),
+    updated_at TIMESTAMP WITH TIME ZONE DEFAULT NOW()
+);
+
+CREATE INDEX IF NOT EXISTS idx_knowledge_embedding
+ON knowledge_chunks USING ivfflat (embedding vector_cosine_ops)
+WITH (lists = 100);
+
+CREATE INDEX IF NOT EXISTS idx_knowledge_category ON knowledge_chunks(category);
+CREATE INDEX IF NOT EXISTS idx_knowledge_source ON knowledge_chunks(source);
+"""
--- a/src/infrastructure/db.py
+++ b/src/infrastructure/db.py
@@ -0,0 +1,62 @@
+"""Database configuration with async SQLAlchemy + pgvector."""
+
+from typing import AsyncGenerator
+
+from sqlalchemy.ext.asyncio import AsyncSession, async_sessionmaker, create_async_engine
+from sqlalchemy.orm import declarative_base
+from sqlalchemy.pool import NullPool
+
+from src.config import settings
+
+# Base class for SQLModel/SQLAlchemy models
+Base = declarative_base()
+
+# Engine configuration
+engine_kwargs = {
+    "pool_size": settings.DATABASE_POOL_SIZE,
+    "max_overflow": settings.DATABASE_MAX_OVERFLOW,
+    "pool_pre_ping": True,
+    "pool_recycle": 300,
+    "echo": settings.DEBUG,
+}
+
+if settings.APP_ENV == "testing":
+    engine_kwargs["poolclass"] = NullPool
+
+engine = create_async_engine(
+    settings.database_url_str,
+    **engine_kwargs,
+)
+
+# Session factory
+AsyncSessionLocal = async_sessionmaker(
+    engine,
+    class_=AsyncSession,
+    expire_on_commit=False,
+    autoflush=False,
+    autocommit=False,
+)
+
+
+async def get_db() -> AsyncGenerator[AsyncSession, None]:
+    """Yield an async database session for dependency injection."""
+    async with AsyncSessionLocal() as session:
+        try:
+            yield session
+            await session.commit()
+        except Exception:
+            await session.rollback()
+            raise
+        finally:
+            await session.close()
+
+
+async def init_db() -> None:
+    """Initialize database tables (for dev/testing only)."""
+    async with engine.begin() as conn:
+        await conn.run_sync(Base.metadata.create_all)
+
+
+async def close_db() -> None:
+    """Dispose database engine."""
+    await engine.dispose()
--- a/src/infrastructure/erpnext/init.py
+++ b/src/infrastructure/erpnext/init.py
--- a/src/infrastructure/erpnext/client.py
+++ b/src/infrastructure/erpnext/client.py
@@ -0,0 +1,255 @@
+"""ERPNext Frappe REST API client."""
+
+from typing import Any
+
+import httpx
+import structlog
+from tenacity import retry, retry_if_exception_type, stop_after_attempt, wait_exponential
+
+from src.config import settings
+from src.core.exceptions import ERPNextError
+
+logger = structlog.get_logger(__name__)
+
+
+class ERPNextClient:
+    """Async client for ERPNext Frappe REST API."""
+
+    def __init__(self) -> None:
+        self.base_url = settings.ERPNEXT_BASE_URL.rstrip("/")
+        self.api_key = settings.ERPNEXT_API_KEY.get_secret_value()
+        self.api_secret = settings.ERPNEXT_API_SECRET.get_secret_value()
+        self.verify_ssl = settings.ERPNEXT_VERIFY_SSL
+
+        self.headers = {
+            "Authorization": f"token {self.api_key}:{self.api_secret}",
+            "Content-Type": "application/json",
+            "Accept": "application/json",
+        }
+        self.client = httpx.AsyncClient(
+            base_url=self.base_url,
+            headers=self.headers,
+            timeout=httpx.Timeout(30.0, connect=10.0),
+            verify=self.verify_ssl,
+        )
+
+    async def close(self) -> None:
+        """Close HTTP client."""
+        await self.client.aclose()
+
+    @retry(
+        stop=stop_after_attempt(3),
+        wait=wait_exponential(multiplier=1, min=2, max=10),
+        retry=retry_if_exception_type((httpx.HTTPStatusError, httpx.NetworkError)),
+        reraise=True,
+    )
+    async def get_document(
+        self,
+        doctype: str,
+        name: str,
+        fields: list[str] | None = None,
+    ) -> dict[str, Any]:
+        """Get a single document from ERPNext."""
+        params: dict[str, Any] = {}
+        if fields:
+            params["fields"] = str(fields)
+
+        response = await self.client.get(
+            f"/api/resource/{doctype}/{name}",
+            params=params,
+        )
+        response.raise_for_status()
+        data = response.json()
+        return data.get("data", {})
+
+    @retry(
+        stop=stop_after_attempt(3),
+        wait=wait_exponential(multiplier=1, min=2, max=10),
+        retry=retry_if_exception_type((httpx.HTTPStatusError, httpx.NetworkError)),
+        reraise=True,
+    )
+    async def get_list(
+        self,
+        doctype: str,
+        filters: list[list[Any]] | None = None,
+        fields: list[str] | None = None,
+        or_filters: list[list[Any]] | None = None,
+        limit: int = 20,
+        limit_start: int = 0,
+        order_by: str = "modified desc",
+    ) -> list[dict[str, Any]]:
+        """Query a list of documents from ERPNext."""
+        params: dict[str, Any] = {
+            "limit_page_length": limit,
+            "limit_start": limit_start,
+            "order_by": order_by,
+        }
+        if filters:
+            params["filters"] = str(filters)
+        if or_filters:
+            params["or_filters"] = str(or_filters)
+        if fields:
+            params["fields"] = str(fields)
+
+        response = await self.client.get(
+            f"/api/resource/{doctype}",
+            params=params,
+        )
+        response.raise_for_status()
+        data = response.json()
+        return data.get("data", [])
+
+    async def create_document(
+        self,
+        doctype: str,
+        data: dict[str, Any],
+    ) -> dict[str, Any]:
+        """Create a new document in ERPNext."""
+        try:
+            response = await self.client.post(
+                f"/api/resource/{doctype}",
+                json=data,
+            )
+            response.raise_for_status()
+            result = response.json()
+            logger.info(
+                "erpnext_document_created",
+                doctype=doctype,
+                name=result.get("data", {}).get("name"),
+            )
+            return result.get("data", {})
+        except httpx.HTTPStatusError as exc:
+            logger.error(
+                "erpnext_create_failed",
+                doctype=doctype,
+                status=exc.response.status_code,
+                response=exc.response.text,
+            )
+            raise ERPNextError(
+                f"Failed to create {doctype}: {exc.response.text}",
+                status_code=exc.response.status_code,
+            ) from exc
+
+    async def update_document(
+        self,
+        doctype: str,
+        name: str,
+        data: dict[str, Any],
+    ) -> dict[str, Any]:
+        """Update an existing document."""
+        try:
+            response = await self.client.put(
+                f"/api/resource/{doctype}/{name}",
+                json=data,
+            )
+            response.raise_for_status()
+            result = response.json()
+            logger.info("erpnext_document_updated", doctype=doctype, name=name)
+            return result.get("data", {})
+        except httpx.HTTPStatusError as exc:
+            raise ERPNextError(
+                f"Failed to update {doctype}: {exc.response.text}",
+                status_code=exc.response.status_code,
+            ) from exc
+
+    async def call_method(
+        self,
+        method: str,
+        data: dict[str, Any] | None = None,
+    ) -> dict[str, Any]:
+        """Call a Frappe method (whitelisted)."""
+        try:
+            response = await self.client.post(
+                f"/api/method/{method}",
+                json=data or {},
+            )
+            response.raise_for_status()
+            return response.json()
+        except httpx.HTTPStatusError as exc:
+            raise ERPNextError(
+                f"Method {method} failed: {exc.response.text}",
+                status_code=exc.response.status_code,
+            ) from exc
+
+    # -------------------------------------------------------------------------
+    # Convenience methods for Healthcare
+    # -------------------------------------------------------------------------
+
+    async def find_patient_by_phone(self, phone: str) -> dict[str, Any] | None:
+        """Find a patient by mobile number."""
+        patients = await self.get_list(
+            "Patient",
+            filters=[["mobile", "=", phone]],
+            fields=["name", "patient_name", "mobile", "sex", "dob", "blood_group"],
+            limit=1,
+        )
+        return patients[0] if patients else None
+
+    async def get_appointments(
+        self,
+        patient: str | None = None,
+        date: str | None = None,
+        status: str | None = None,
+    ) -> list[dict[str, Any]]:
+        """Get patient appointments."""
+        filters: list[list[Any]] = []
+        if patient:
+            filters.append(["patient", "=", patient])
+        if date:
+            filters.append(["appointment_date", "=", date])
+        if status:
+            filters.append(["status", "=", status])
+
+        return await self.get_list(
+            "Patient Appointment",
+            filters=filters if filters else None,
+            fields=[
+                "name", "patient", "practitioner", "appointment_date",
+                "appointment_time", "duration", "status", "department",
+            ],
+            limit=50,
+        )
+
+    async def create_appointment(
+        self,
+        patient: str,
+        practitioner: str,
+        appointment_date: str,
+        appointment_time: str,
+        duration: int = 30,
+        department: str = "Dermatología Estética",
+        notes: str = "",
+    ) -> dict[str, Any]:
+        """Create a patient appointment."""
+        data = {
+            "doctype": "Patient Appointment",
+            "patient": patient,
+            "practitioner": practitioner,
+            "appointment_date": appointment_date,
+            "appointment_time": appointment_time,
+            "duration": duration,
+            "department": department,
+            "notes": notes,
+            "status": "Scheduled",
+        }
+        return await self.create_document("Patient Appointment", data)
+
+
+# Global singleton
+_erpnext_client: ERPNextClient | None = None
+
+
+async def get_erpnext_client() -> ERPNextClient:
+    """Return ERPNext client singleton."""
+    global _erpnext_client
+    if _erpnext_client is None:
+        _erpnext_client = ERPNextClient()
+    return _erpnext_client
+
+
+async def close_erpnext_client() -> None:
+    """Close ERPNext client."""
+    global _erpnext_client
+    if _erpnext_client:
+        await _erpnext_client.close()
+        _erpnext_client = None
--- a/src/infrastructure/redis.py
+++ b/src/infrastructure/redis.py
@@ -0,0 +1,69 @@
+"""Redis client configuration."""
+
+import orjson
+import redis.asyncio as aioredis
+from redis.asyncio import Redis
+
+from src.config import settings
+
+# Global Redis client instance
+_redis_client: Redis | None = None
+
+
+async def get_redis() -> Redis:
+    """Return async Redis client singleton."""
+    global _redis_client
+    if _redis_client is None:
+        _redis_client = aioredis.from_url(
+            str(settings.REDIS_URL),
+            decode_responses=True,
+            encoding="utf-8",
+        )
+    return _redis_client
+
+
+async def close_redis() -> None:
+    """Close Redis connection."""
+    global _redis_client
+    if _redis_client:
+        await _redis_client.close()
+        _redis_client = None
+
+
+class RedisCache:
+    """Helper class for common Redis operations."""
+
+    def __init__(self, redis: Redis) -> None:
+        self.redis = redis
+
+    async def get(self, key: str) -> dict | None:
+        """Get and deserialize JSON value."""
+        value = await self.redis.get(key)
+        if value is None:
+            return None
+        return orjson.loads(value)
+
+    async def set(
+        self,
+        key: str,
+        value: dict,
+        ttl: int = 3600,
+    ) -> None:
+        """Serialize and set JSON value with TTL."""
+        await self.redis.setex(key, ttl, orjson.dumps(value))
+
+    async def delete(self, key: str) -> None:
+        """Delete key."""
+        await self.redis.delete(key)
+
+    async def exists(self, key: str) -> bool:
+        """Check if key exists."""
+        return await self.redis.exists(key) > 0
+
+    async def increment(self, key: str, amount: int = 1) -> int:
+        """Increment counter."""
+        return await self.redis.incrby(key, amount)
+
+    async def expire(self, key: str, ttl: int) -> None:
+        """Set expiration on key."""
+        await self.redis.expire(key, ttl)
--- a/src/infrastructure/whatsapp/init.py
+++ b/src/infrastructure/whatsapp/init.py
--- a/src/infrastructure/whatsapp/client.py
+++ b/src/infrastructure/whatsapp/client.py
@@ -0,0 +1,254 @@
+"""Official Meta WhatsApp Business API client."""
+
+import asyncio
+from typing import Any
+
+import httpx
+import structlog
+from tenacity import retry, retry_if_exception_type, stop_after_attempt, wait_exponential
+
+from src.config import settings
+from src.core.exceptions import WhatsAppAPIError
+
+logger = structlog.get_logger(__name__)
+
+
+class WhatsAppClient:
+    """Async client for Meta WhatsApp Business API."""
+
+    def __init__(self) -> None:
+        self.base_url = settings.meta_api_base_url
+        self.phone_number_id = settings.META_PHONE_NUMBER_ID
+        self.access_token = settings.META_ACCESS_TOKEN.get_secret_value()
+        self.headers = {
+            "Authorization": f"Bearer {self.access_token}",
+            "Content-Type": "application/json",
+        }
+        self.client = httpx.AsyncClient(
+            base_url=self.base_url,
+            headers=self.headers,
+            timeout=httpx.Timeout(30.0, connect=10.0),
+            http2=True,
+        )
+
+    async def close(self) -> None:
+        """Close HTTP client."""
+        await self.client.aclose()
+
+    @retry(
+        stop=stop_after_attempt(3),
+        wait=wait_exponential(multiplier=1, min=2, max=10),
+        retry=retry_if_exception_type((httpx.HTTPStatusError, httpx.NetworkError)),
+        reraise=True,
+    )
+    async def _post(self, endpoint: str, payload: dict[str, Any]) -> dict[str, Any]:
+        """Make POST request with retries."""
+        response = await self.client.post(endpoint, json=payload)
+        response.raise_for_status()
+        return response.json()
+
+    async def send_text_message(
+        self,
+        to: str,
+        text: str,
+        preview_url: bool = False,
+    ) -> dict[str, Any]:
+        """Send a text message to a WhatsApp user.
+
+        Args:
+            to: Recipient phone number in international format (e.g., 5216641234567).
+            text: Message body (max 4096 chars).
+            preview_url: Whether to show URL preview.
+
+        Returns:
+            API response with message ID.
+        """
+        if len(text) > 4096:
+            text = text[:4093] + "..."
+
+        payload = {
+            "messaging_product": "whatsapp",
+            "recipient_type": "individual",
+            "to": to,
+            "type": "text",
+            "text": {"preview_url": preview_url, "body": text},
+        }
+
+        endpoint = f"/{self.phone_number_id}/messages"
+        try:
+            result = await self._post(endpoint, payload)
+            logger.info(
+                "whatsapp_text_sent",
+                to=to,
+                message_id=result.get("messages", [{}])[0].get("id"),
+            )
+            return result
+        except httpx.HTTPStatusError as exc:
+            logger.error(
+                "whatsapp_text_failed",
+                to=to,
+                status_code=exc.response.status_code,
+                response=exc.response.text,
+            )
+            raise WhatsAppAPIError(
+                f"Failed to send WhatsApp message: {exc.response.text}",
+                status_code=exc.response.status_code,
+            ) from exc
+
+    async def send_template_message(
+        self,
+        to: str,
+        template_name: str,
+        language_code: str = "es_MX",
+        components: list[dict[str, Any]] | None = None,
+    ) -> dict[str, Any]:
+        """Send a template message (required for 24h+ inactive conversations)."""
+        payload: dict[str, Any] = {
+            "messaging_product": "whatsapp",
+            "recipient_type": "individual",
+            "to": to,
+            "type": "template",
+            "template": {
+                "name": template_name,
+                "language": {"code": language_code},
+            },
+        }
+        if components:
+            payload["template"]["components"] = components
+
+        endpoint = f"/{self.phone_number_id}/messages"
+        try:
+            result = await self._post(endpoint, payload)
+            logger.info(
+                "whatsapp_template_sent",
+                to=to,
+                template=template_name,
+                message_id=result.get("messages", [{}])[0].get("id"),
+            )
+            return result
+        except httpx.HTTPStatusError as exc:
+            raise WhatsAppAPIError(
+                f"Failed to send template: {exc.response.text}",
+                status_code=exc.response.status_code,
+            ) from exc
+
+    async def send_interactive_buttons(
+        self,
+        to: str,
+        body: str,
+        buttons: list[dict[str, str]],
+    ) -> dict[str, Any]:
+        """Send interactive button message.
+
+        Args:
+            buttons: List of {"id": "btn_1", "title": "Option 1"} (max 3).
+        """
+        if len(buttons) > 3:
+            raise ValueError("Maximum 3 buttons allowed")
+
+        payload = {
+            "messaging_product": "whatsapp",
+            "recipient_type": "individual",
+            "to": to,
+            "type": "interactive",
+            "interactive": {
+                "type": "button",
+                "body": {"text": body},
+                "action": {
+                    "buttons": [
+                        {"type": "reply", "reply": {"id": b["id"], "title": b["title"]}}
+                        for b in buttons
+                    ]
+                },
+            },
+        }
+
+        endpoint = f"/{self.phone_number_id}/messages"
+        try:
+            result = await self._post(endpoint, payload)
+            logger.info("whatsapp_buttons_sent", to=to)
+            return result
+        except httpx.HTTPStatusError as exc:
+            raise WhatsAppAPIError(
+                f"Failed to send buttons: {exc.response.text}",
+                status_code=exc.response.status_code,
+            ) from exc
+
+    async def send_interactive_list(
+        self,
+        to: str,
+        body: str,
+        button_text: str,
+        sections: list[dict[str, Any]],
+    ) -> dict[str, Any]:
+        """Send interactive list message (e.g., service selection)."""
+        payload = {
+            "messaging_product": "whatsapp",
+            "recipient_type": "individual",
+            "to": to,
+            "type": "interactive",
+            "interactive": {
+                "type": "list",
+                "body": {"text": body},
+                "action": {
+                    "button": button_text,
+                    "sections": sections,
+                },
+            },
+        }
+
+        endpoint = f"/{self.phone_number_id}/messages"
+        try:
+            result = await self._post(endpoint, payload)
+            logger.info("whatsapp_list_sent", to=to)
+            return result
+        except httpx.HTTPStatusError as exc:
+            raise WhatsAppAPIError(
+                f"Failed to send list: {exc.response.text}",
+                status_code=exc.response.status_code,
+            ) from exc
+
+    async def mark_as_read(self, message_id: str) -> dict[str, Any]:
+        """Mark a message as read."""
+        payload = {
+            "messaging_product": "whatsapp",
+            "status": "read",
+            "message_id": message_id,
+        }
+        endpoint = f"/{self.phone_number_id}/messages"
+        try:
+            result = await self._post(endpoint, payload)
+            logger.info("whatsapp_marked_read", message_id=message_id)
+            return result
+        except httpx.HTTPStatusError as exc:
+            raise WhatsAppAPIError(
+                f"Failed to mark as read: {exc.response.text}",
+                status_code=exc.response.status_code,
+            ) from exc
+
+    async def get_business_profile(self) -> dict[str, Any]:
+        """Get business profile information."""
+        endpoint = f"/{self.phone_number_id}/whatsapp_business_profile"
+        response = await self.client.get(endpoint, params={"fields": "about,description,email"})
+        response.raise_for_status()
+        return response.json()
+
+
+# Global client instance
+_whatsapp_client: WhatsAppClient | None = None
+
+
+async def get_whatsapp_client() -> WhatsAppClient:
+    """Return WhatsApp client singleton."""
+    global _whatsapp_client
+    if _whatsapp_client is None:
+        _whatsapp_client = WhatsAppClient()
+    return _whatsapp_client
+
+
+async def close_whatsapp_client() -> None:
+    """Close WhatsApp client."""
+    global _whatsapp_client
+    if _whatsapp_client:
+        await _whatsapp_client.close()
+        _whatsapp_client = None
--- a/src/infrastructure/whatsapp/webhook.py
+++ b/src/infrastructure/whatsapp/webhook.py
@@ -0,0 +1,168 @@
+"""Meta WhatsApp webhook parser and validator."""
+
+import hmac
+import hashlib
+from typing import Any
+
+import structlog
+
+from src.config import settings
+from src.core.constants import WhatsAppMessageType
+from src.core.exceptions import ValidationError
+
+logger = structlog.get_logger(__name__)
+
+
+class WhatsAppWebhookPayload:
+    """Parsed WhatsApp webhook payload."""
+
+    def __init__(self, raw: dict[str, Any]) -> None:
+        self.raw = raw
+        self._entry = raw.get("entry", [{}])[0]
+        self._change = self._entry.get("changes", [{}])[0]
+        self._value = self._change.get("value", {})
+        self._message = self._value.get("messages", [{}])[0] if "messages" in self._value else {}
+
+    @property
+    def object_type(self) -> str:
+        return self.raw.get("object", "")
+
+    @property
+    def business_phone_number_id(self) -> str:
+        return self._value.get("metadata", {}).get("phone_number_id", "")
+
+    @property
+    def display_phone_number(self) -> str:
+        return self._value.get("metadata", {}).get("display_phone_number", "")
+
+    @property
+    def has_messages(self) -> bool:
+        return "messages" in self._value and len(self._value["messages"]) > 0
+
+    @property
+    def has_statuses(self) -> bool:
+        return "statuses" in self._value and len(self._value["statuses"]) > 0
+
+    # Message properties
+    @property
+    def message_id(self) -> str:
+        return self._message.get("id", "")
+
+    @property
+    def from_number(self) -> str:
+        return self._message.get("from", "")
+
+    @property
+    def timestamp(self) -> str:
+        return self._message.get("timestamp", "")
+
+    @property
+    def message_type(self) -> WhatsAppMessageType:
+        msg_type = self._message.get("type", "")
+        try:
+            return WhatsAppMessageType(msg_type)
+        except ValueError:
+            return WhatsAppMessageType.UNKNOWN
+
+    @property
+    def text_body(self) -> str:
+        if self.message_type == WhatsAppMessageType.TEXT:
+            return self._message.get("text", {}).get("body", "")
+        return ""
+
+    @property
+    def button_payload(self) -> str:
+        if self.message_type == WhatsAppMessageType.INTERACTIVE:
+            interactive = self._message.get("interactive", {})
+            if "button_reply" in interactive:
+                return interactive["button_reply"].get("id", "")
+            if "list_reply" in interactive:
+                return interactive["list_reply"].get("id", "")
+        return ""
+
+    @property
+    def button_title(self) -> str:
+        if self.message_type == WhatsAppMessageType.INTERACTIVE:
+            interactive = self._message.get("interactive", {})
+            if "button_reply" in interactive:
+                return interactive["button_reply"].get("title", "")
+            if "list_reply" in interactive:
+                return interactive["list_reply"].get("title", "")
+        return ""
+
+    @property
+    def image_data(self) -> dict[str, Any]:
+        if self.message_type == WhatsAppMessageType.IMAGE:
+            return self._message.get("image", {})
+        return {}
+
+    @property
+    def audio_data(self) -> dict[str, Any]:
+        if self.message_type == WhatsAppMessageType.AUDIO:
+            return self._message.get("audio", {})
+        return {}
+
+    @property
+    def context(self) -> dict[str, Any]:
+        """Context of a reply (original message ID)."""
+        return self._message.get("context", {})
+
+    @property
+    def is_reply(self) -> bool:
+        return bool(self.context.get("id"))
+
+    # Status properties
+    @property
+    def statuses(self) -> list[dict[str, Any]]:
+        return self._value.get("statuses", [])
+
+    def __repr__(self) -> str:
+        return (
+            f"<WhatsAppWebhookPayload from={self.from_number} "
+            f"type={self.message_type.value} id={self.message_id}>"
+        )
+
+
+class WebhookVerifier:
+    """Verify Meta webhook signatures."""
+
+    @staticmethod
+    def verify_subscription(mode: str | None, token: str | None) -> bool:
+        """Verify webhook subscription challenge."""
+        if mode != "subscribe":
+            return False
+        verify_token = settings.META_WEBHOOK_VERIFY_TOKEN.get_secret_value()
+        return token == verify_token
+
+    @staticmethod
+    def verify_signature(body: bytes, signature: str | None) -> bool:
+        """Verify X-Hub-Signature-256 header."""
+        if not signature:
+            logger.warning("missing_webhook_signature")
+            return False
+
+        app_secret = settings.META_APP_SECRET.get_secret_value()
+        expected = hmac.new(
+            app_secret.encode("utf-8"),
+            body,
+            hashlib.sha256,
+        ).hexdigest()
+
+        if not hmac.compare_digest(f"sha256={expected}", signature):
+            logger.warning("invalid_webhook_signature")
+            return False
+
+        return True
+
+
+def parse_webhook_payload(payload: dict[str, Any]) -> WhatsAppWebhookPayload:
+    """Parse and validate incoming webhook payload."""
+    if payload.get("object") != "whatsapp_business_account":
+        raise ValidationError("Invalid webhook object type")
+
+    parsed = WhatsAppWebhookPayload(payload)
+
+    if not parsed.has_messages and not parsed.has_statuses:
+        raise ValidationError("Webhook contains no messages or statuses")
+
+    return parsed
--- a/src/main.py
+++ b/src/main.py
@@ -0,0 +1,170 @@
+"""SKEEN CRM AI Agent - FastAPI Application Entry Point."""
+
+import time
+import uuid
+from contextlib import asynccontextmanager
+
+import structlog
+from asgi_correlation_id import CorrelationIdMiddleware
+from fastapi import FastAPI, Request, status
+from fastapi.middleware.cors import CORSMiddleware
+from fastapi.responses import JSONResponse
+from prometheus_client import make_asgi_app
+
+from src.api.v1.health import router as health_router
+from src.api.v1.messages import router as messages_router
+from src.api.v1.webhooks import router as webhooks_router
+from src.config import settings
+from src.core.exceptions import SkeenException
+from src.infrastructure.db import close_db, init_db
+from src.infrastructure.erpnext.client import close_erpnext_client
+from src.infrastructure.redis import close_redis
+from src.infrastructure.whatsapp.client import close_whatsapp_client
+
+logger = structlog.get_logger(__name__)
+
+
+def setup_logging() -> None:
+    """Configure structured logging."""
+    structlog.configure(
+        processors=[
+            structlog.stdlib.filter_by_level,
+            structlog.stdlib.add_logger_name,
+            structlog.stdlib.add_log_level,
+            structlog.stdlib.PositionalArgumentsFormatter(),
+            structlog.processors.TimeStamper(fmt="iso"),
+            structlog.processors.StackInfoRenderer(),
+            structlog.processors.format_exc_info,
+            structlog.processors.UnicodeDecoder(),
+            structlog.processors.JSONRenderer(),
+        ],
+        context_class=dict,
+        logger_factory=structlog.stdlib.LoggerFactory(),
+        wrapper_class=structlog.stdlib.BoundLogger,
+        cache_logger_on_first_use=True,
+    )
+
+
+@asynccontextmanager
+async def lifespan(app: FastAPI):
+    """Application lifespan events."""
+    setup_logging()
+    logger.info(
+        "application_starting",
+        app_name=settings.APP_NAME,
+        environment=settings.APP_ENV,
+        version="0.1.0",
+    )
+
+    # Initialize database in development
+    if settings.APP_ENV == "development":
+        try:
+            await init_db()
+            logger.info("database_initialized")
+        except Exception as exc:
+            logger.warning("database_init_skipped", error=str(exc))
+
+    yield
+
+    # Cleanup
+    logger.info("application_shutting_down")
+    await close_redis()
+    await close_whatsapp_client()
+    await close_erpnext_client()
+    await close_db()
+    logger.info("application_shutdown_complete")
+
+
+def create_app() -> FastAPI:
+    """Application factory."""
+    app = FastAPI(
+        title=settings.APP_NAME,
+        description="SKEEN CRM AI Agent for WhatsApp Business API + ERPNext",
+        version="0.1.0",
+        docs_url="/docs" if not settings.is_production else None,
+        redoc_url="/redoc" if not settings.is_production else None,
+        openapi_url="/openapi.json" if not settings.is_production else None,
+        lifespan=lifespan,
+    )
+
+    # Middleware
+    app.add_middleware(
+        CORSMiddleware,
+        allow_origins=["*"],  # TODO: Restrict in production
+        allow_credentials=True,
+        allow_methods=["*"],
+        allow_headers=["*"],
+    )
+    app.add_middleware(CorrelationIdMiddleware)
+
+    # Request timing middleware
+    @app.middleware("http")
+    async def add_process_time_header(request: Request, call_next):
+        start_time = time.time()
+        request_id = str(uuid.uuid4())
+        structlog.contextvars.clear_contextvars()
+        structlog.contextvars.bind_contextvars(
+            request_id=request_id,
+            path=request.url.path,
+            method=request.method,
+        )
+
+        try:
+            response = await call_next(request)
+            process_time = time.time() - start_time
+            response.headers["X-Process-Time"] = str(process_time)
+            response.headers["X-Request-ID"] = request_id
+
+            logger.info(
+                "request_completed",
+                status_code=response.status_code,
+                duration_ms=round(process_time * 1000, 2),
+            )
+            return response
+        except Exception as exc:
+            logger.error("request_failed", error=str(exc))
+            raise
+
+    # Exception handlers
+    @app.exception_handler(SkeenException)
+    async def skeen_exception_handler(request: Request, exc: SkeenException):
+        logger.error(
+            "application_exception",
+            error=exc.message,
+            status_code=exc.status_code,
+            details=exc.details,
+        )
+        return JSONResponse(
+            status_code=exc.status_code,
+            content={
+                "error": exc.message,
+                "details": exc.details,
+                "request_id": str(uuid.uuid4()),
+            },
+        )
+
+    @app.exception_handler(Exception)
+    async def generic_exception_handler(request: Request, exc: Exception):
+        logger.exception("unhandled_exception", error=str(exc))
+        return JSONResponse(
+            status_code=status.HTTP_500_INTERNAL_SERVER_ERROR,
+            content={
+                "error": "Internal server error",
+                "request_id": str(uuid.uuid4()),
+            },
+        )
+
+    # Routes
+    app.include_router(health_router, prefix="/api/v1")
+    app.include_router(webhooks_router, prefix="/api/v1")
+    app.include_router(messages_router, prefix="/api/v1")
+
+    # Metrics endpoint (Prometheus)
+    if settings.ENABLE_METRICS:
+        metrics_app = make_asgi_app()
+        app.mount("/metrics", metrics_app)
+
+    return app
+
+
+app = create_app()
--- a/src/use_cases/init.py
+++ b/src/use_cases/init.py
--- a/src/use_cases/handle_incoming_message.py
+++ b/src/use_cases/handle_incoming_message.py
@@ -0,0 +1,372 @@
+"""Core use case: process incoming WhatsApp message with AI agent."""
+
+import json
+import uuid
+from datetime import datetime, timezone
+from typing import Any
+
+import structlog
+from sqlalchemy import select
+from sqlalchemy.ext.asyncio import AsyncSession
+
+from src.config import settings
+from src.core.constants import ConversationStatus, SKEEN_SYSTEM_PROMPT, WhatsAppMessageType
+from src.infrastructure.ai.openai_client import get_openai_client
+from src.infrastructure.ai.prompts import TOOLS
+from src.infrastructure.ai.rag import RAGStore
+from src.infrastructure.erpnext.client import get_erpnext_client
+from src.infrastructure.whatsapp.client import get_whatsapp_client
+from src.infrastructure.whatsapp.webhook import WhatsAppWebhookPayload
+from src.domain.models.conversation import Conversation, Message
+
+logger = structlog.get_logger(__name__)
+
+MAX_CONTEXT_MESSAGES = 10
+
+
+class ToolExecutor:
+    """Executes tool calls requested by the LLM."""
+
+    def __init__(self, session: AsyncSession) -> None:
+        self.session = session
+        self.rag = RAGStore(session)
+        self.erpnext = None  # Lazy init
+
+    async def execute(self, tool_call: dict[str, Any]) -> dict[str, Any]:
+        """Execute a single tool call and return result."""
+        name = tool_call["function"]["name"]
+        arguments = json.loads(tool_call["function"]["arguments"])
+        tool_call_id = tool_call["id"]
+
+        logger.info("executing_tool", tool=name, args=arguments)
+
+        try:
+            if name == "search_catalog":
+                result = await self._search_catalog(arguments)
+            elif name == "check_availability":
+                result = await self._check_availability(arguments)
+            elif name == "create_appointment":
+                result = await self._create_appointment(arguments)
+            elif name == "get_patient_info":
+                result = await self._get_patient_info(arguments)
+            elif name == "get_wallet_balance":
+                result = await self._get_wallet_balance(arguments)
+            elif name == "escalate_to_human":
+                result = await self._escalate_to_human(arguments)
+            else:
+                result = {"error": f"Unknown tool: {name}"}
+        except Exception as exc:
+            logger.error("tool_execution_failed", tool=name, error=str(exc))
+            result = {"error": f"Failed to execute {name}: {str(exc)}"}
+
+        return {
+            "tool_call_id": tool_call_id,
+            "role": "tool",
+            "name": name,
+            "content": json.dumps(result, ensure_ascii=False),
+        }
+
+    async def _search_catalog(self, args: dict[str, Any]) -> dict[str, Any]:
+        query = args.get("query", "")
+        category = args.get("category")
+        results = await self.rag.search(query, category=category, top_k=5)
+        return {
+            "results": [
+                {
+                    "content": r["content"],
+                    "category": r["category"],
+                    "source": r["source"],
+                    "similarity": round(r["similarity"], 3),
+                }
+                for r in results
+            ]
+        }
+
+    async def _check_availability(self, args: dict[str, Any]) -> dict[str, Any]:
+        date = args.get("date")
+        branch = args.get("branch")
+        doctor = args.get("doctor")
+        service = args.get("service")
+
+        # TODO: Integrate with ERPNext Healthcare scheduling
+        # For now, return mock data structure
+        return {
+            "date": date,
+            "available_slots": [
+                {"time": "10:00", "doctor": "Dr. Ramos"},
+                {"time": "11:30", "doctor": "Dr. Martínez"},
+                {"time": "15:00", "doctor": "Dr. Ramos"},
+            ],
+            "branch": branch or "Rosarito",
+            "service": service,
+            "note": "Esta es una respuesta simulada. Integrar con ERPNext Healthcare.",
+        }
+
+    async def _create_appointment(self, args: dict[str, Any]) -> dict[str, Any]:
+        # TODO: Integrate with ERPNext to create real appointments
+        return {
+            "status": "simulated",
+            "appointment_id": f"APT-{uuid.uuid4().hex[:8].upper()}",
+            "details": args,
+            "note": "Cita simulada. Integrar con ERPNext Patient Appointment.",
+        }
+
+    async def _get_patient_info(self, args: dict[str, Any]) -> dict[str, Any]:
+        phone = args.get("phone", "")
+        erpnext = await get_erpnext_client()
+        patient = await erpnext.find_patient_by_phone(phone)
+
+        if not patient:
+            return {"found": False, "message": "No se encontró paciente con ese número."}
+
+        appointments = await erpnext.get_appointments(patient=patient.get("name"))
+        return {
+            "found": True,
+            "name": patient.get("patient_name"),
+            "sex": patient.get("sex"),
+            "blood_group": patient.get("blood_group"),
+            "total_appointments": len(appointments),
+            "last_appointments": appointments[:3],
+        }
+
+    async def _get_wallet_balance(self, args: dict[str, Any]) -> dict[str, Any]:
+        # TODO: Integrate with ERPNext custom Wallet doctype
+        return {
+            "balance_mxn": 0.0,
+            "points": 0,
+            "note": "Monedero no implementado en ERPNext aún.",
+        }
+
+    async def _escalate_to_human(self, args: dict[str, Any]) -> dict[str, Any]:
+        reason = args.get("reason", "Solicitud del paciente")
+        return {
+            "escalated": True,
+            "reason": reason,
+            "message": "Un agente humano de SKEEN se pondrá en contacto contigo pronto. ⏳",
+        }
+
+
+async def get_or_create_conversation(
+    db: AsyncSession,
+    phone_number: str,
+) -> Conversation:
+    """Get existing conversation or create new one."""
+    result = await db.execute(
+        select(Conversation)
+        .where(Conversation.phone_number == phone_number)
+        .order_by(Conversation.created_at.desc())
+    )
+    conversation = result.scalars().first()
+
+    if conversation is None:
+        conversation = Conversation(
+            id=str(uuid.uuid4()),
+            phone_number=phone_number,
+            status=ConversationStatus.ACTIVE,
+            context={},
+        )
+        db.add(conversation)
+        await db.flush()
+        logger.info("new_conversation_created", phone=phone_number, conversation_id=conversation.id)
+    else:
+        # Reactivate if resolved
+        if conversation.status == ConversationStatus.RESOLVED:
+            conversation.status = ConversationStatus.ACTIVE
+        conversation.last_message_at = datetime.now(timezone.utc)
+
+    return conversation
+
+
+async def get_conversation_history(
+    db: AsyncSession,
+    conversation_id: str,
+    limit: int = MAX_CONTEXT_MESSAGES,
+) -> list[dict[str, str]]:
+    """Get recent messages formatted for OpenAI context."""
+    from sqlalchemy import select
+    from src.domain.models.conversation import Message
+
+    result = await db.execute(
+        select(Message)
+        .where(Message.conversation_id == conversation_id)
+        .order_by(Message.created_at.desc())
+        .limit(limit)
+    )
+    messages = result.scalars().all()
+
+    # Reverse to chronological order
+    history = []
+    for msg in reversed(messages):
+        history.append({"role": msg.role, "content": msg.content})
+    return history
+
+
+async def process_incoming_message(
+    db: AsyncSession,
+    webhook_data: dict[str, Any],
+    client_ip: str = "unknown",
+) -> dict[str, Any]:
+    """Process an incoming WhatsApp message end-to-end.
+
+    1. Parse webhook
+    2. Load/create conversation
+    3. Build LLM context
+    4. Call OpenAI with tools
+    5. Execute tools if needed
+    6. Send response
+    7. Persist everything
+    """
+    webhook = WhatsAppWebhookPayload(webhook_data)
+
+    if not webhook.has_messages:
+        return {"status": "no_messages"}
+
+    phone = webhook.from_number
+    message_text = webhook.text_body or "[Mensaje no textual]"
+    message_type = webhook.message_type.value
+
+    logger.info(
+        "incoming_message_received",
+        from_number=phone[-4:],  # Log last 4 digits only for privacy
+        message_type=message_type,
+        ip=client_ip,
+    )
+
+    # Get or create conversation
+    conversation = await get_or_create_conversation(db, phone)
+
+    # Try to find patient in ERPNext for personalization
+    patient_name = None
+    try:
+        erpnext = await get_erpnext_client()
+        patient = await erpnext.find_patient_by_phone(phone)
+        if patient:
+            patient_name = patient.get("patient_name")
+            conversation.patient_id = patient.get("name")
+            conversation.patient_name = patient_name
+    except Exception as exc:
+        logger.warning("patient_lookup_failed", error=str(exc))
+
+    # Save inbound message
+    inbound_msg = Message(
+        id=str(uuid.uuid4()),
+        conversation_id=conversation.id,
+        direction="inbound",
+        role="user",
+        message_type=message_type,
+        content=message_text,
+        whatsapp_message_id=webhook.message_id,
+        metadata={"ip": client_ip, "timestamp": webhook.timestamp},
+    )
+    db.add(inbound_msg)
+
+    # Build context for OpenAI
+    system_prompt = SKEEN_SYSTEM_PROMPT
+    if patient_name:
+        system_prompt += f"\n\nPACIENTE ACTUAL: {patient_name}. Salúdalo/a por su nombre cuando sea apropiado."
+
+    messages: list[dict[str, str]] = [
+        {"role": "system", "content": system_prompt},
+    ]
+
+    # Add conversation history
+    history = await get_conversation_history(db, conversation.id)
+    messages.extend(history)
+
+    # Add current message
+    messages.append({"role": "user", "content": message_text})
+
+    # Call OpenAI
+    openai_client = await get_openai_client()
+    llm_response = await openai_client.chat_completion(
+        messages=messages,
+        tools=TOOLS,
+    )
+
+    assistant_message = llm_response["content"] or ""
+    tool_calls = llm_response.get("tool_calls")
+    tool_results = []
+
+    # Execute tools if requested
+    if tool_calls:
+        executor = ToolExecutor(db)
+        for tc in tool_calls:
+            result = await executor.execute(tc)
+            tool_results.append(result)
+
+        # Second LLM call with tool results
+        messages.append({
+            "role": "assistant",
+            "content": assistant_message,
+            "tool_calls": tool_calls,
+        })
+        for tr in tool_results:
+            messages.append({
+                "role": "tool",
+                "tool_call_id": tr["tool_call_id"],
+                "name": tr["name"],
+                "content": tr["content"],
+            })
+
+        final_response = await openai_client.chat_completion(
+            messages=messages,
+            tools=TOOLS,
+        )
+        assistant_message = final_response["content"] or assistant_message
+
+    # Mark as read (best effort)
+    try:
+        wa_client = await get_whatsapp_client()
+        await wa_client.mark_as_read(webhook.message_id)
+    except Exception:
+        pass
+
+    # Send response
+    if assistant_message:
+        try:
+            wa_client = await get_whatsapp_client()
+            send_result = await wa_client.send_text_message(phone, assistant_message)
+            response_message_id = send_result.get("messages", [{}])[0].get("id")
+        except Exception as exc:
+            logger.error("failed_to_send_response", error=str(exc))
+            response_message_id = None
+    else:
+        response_message_id = None
+
+    # Save outbound message
+    outbound_msg = Message(
+        id=str(uuid.uuid4()),
+        conversation_id=conversation.id,
+        direction="outbound",
+        role="assistant",
+        message_type="text",
+        content=assistant_message,
+        whatsapp_message_id=response_message_id,
+        tool_calls=tool_calls,
+        tool_results=tool_results,
+        tokens_used=llm_response.get("usage", {}).get("total_tokens", 0),
+        metadata={"model": settings.OPENAI_MODEL},
+    )
+    db.add(outbound_msg)
+
+    # Update conversation
+    conversation.last_message_at = datetime.now(timezone.utc)
+    if any(tr.get("name") == "escalate_to_human" for tr in tool_results):
+        conversation.status = ConversationStatus.ESCALATED
+
+    await db.commit()
+
+    logger.info(
+        "message_processed",
+        conversation_id=conversation.id,
+        patient=patient_name,
+        tools_used=len(tool_calls) if tool_calls else 0,
+        response_length=len(assistant_message),
+    )
+
+    return {
+        "status": "processed",
+        "conversation_id": conversation.id,
+        "response": assistant_message,
+        "tools_executed": [tr["name"] for tr in tool_results],
+    }
--- a/src/workers/init.py
+++ b/src/workers/init.py
--- a/src/workers/celery_app.py
+++ b/src/workers/celery_app.py
@@ -0,0 +1,61 @@
+"""Celery configuration for background task processing."""
+
+import os
+
+from celery import Celery
+from celery.signals import setup_logging
+
+from src.config import settings
+
+os.environ.setdefault("CELERY_CONFIG_MODULE", "src.workers.celery_app")
+
+celery_app = Celery(
+    "skeen_crm",
+    broker=str(settings.CELERY_BROKER_URL),
+    backend=str(settings.CELERY_RESULT_BACKEND),
+    include=["src.workers.tasks"],
+)
+
+celery_app.conf.update(
+    task_serializer="json",
+    accept_content=["json"],
+    result_serializer="json",
+    timezone="America/Tijuana",
+    enable_utc=True,
+    task_track_started=True,
+    task_time_limit=300,
+    worker_prefetch_multiplier=1,
+    worker_concurrency=settings.CELERY_WORKER_CONCURRENCY,
+    task_routes={
+        "src.workers.tasks.process_whatsapp_message_task": {"queue": "whatsapp"},
+        "src.workers.tasks.sync_erpnext_task": {"queue": "erpnext"},
+        "src.workers.tasks.generate_embedding_task": {"queue": "ai"},
+    },
+    task_default_queue="default",
+)
+
+
+@setup_logging.connect
+def config_loggers(*args, **kwargs) -> None:
+    """Configure structlog for Celery workers."""
+    import logging
+    import structlog
+
+    structlog.configure(
+        processors=[
+            structlog.stdlib.filter_by_level,
+            structlog.stdlib.add_logger_name,
+            structlog.stdlib.add_log_level,
+            structlog.stdlib.PositionalArgumentsFormatter(),
+            structlog.processors.TimeStamper(fmt="iso"),
+            structlog.processors.StackInfoRenderer(),
+            structlog.processors.format_exc_info,
+            structlog.processors.UnicodeDecoder(),
+            structlog.processors.JSONRenderer(),
+        ],
+        context_class=dict,
+        logger_factory=structlog.stdlib.LoggerFactory(),
+        wrapper_class=structlog.stdlib.BoundLogger,
+        cache_logger_on_first_use=True,
+    )
+    logging.basicConfig(format="%(message)s", level=logging.INFO)
--- a/src/workers/tasks.py
+++ b/src/workers/tasks.py
@@ -0,0 +1,67 @@
+"""Celery background tasks."""
+
+import structlog
+
+from src.infrastructure.db import AsyncSessionLocal
+from src.use_cases.handle_incoming_message import process_incoming_message
+from src.workers.celery_app import celery_app
+
+logger = structlog.get_logger(__name__)
+
+
+@celery_app.task(bind=True, max_retries=3, default_retry_delay=10)
+def process_whatsapp_message_task(self, webhook_data: dict, client_ip: str = "unknown") -> dict:
+    """Process WhatsApp message asynchronously.
+
+    This task runs in a Celery worker and handles the full AI pipeline
+    so the webhook endpoint can return 200 immediately to Meta.
+    """
+    import asyncio
+
+    async def _process() -> dict:
+        async with AsyncSessionLocal() as db:
+            try:
+                result = await process_incoming_message(
+                    db=db,
+                    webhook_data=webhook_data,
+                    client_ip=client_ip,
+                )
+                return result
+            except Exception as exc:
+                logger.error("whatsapp_task_failed", error=str(exc), attempt=self.request.retries)
+                raise self.retry(exc=exc)
+
+    return asyncio.run(_process())
+
+
+@celery_app.task
+def sync_erpnext_patient_task(patient_data: dict) -> dict:
+    """Sync patient data to ERPNext in background."""
+    logger.info("syncing_patient_to_erpnext", patient=patient_data.get("name"))
+    # TODO: Implement ERPNext sync logic
+    return {"status": "synced", "patient": patient_data.get("name")}
+
+
+@celery_app.task
+def generate_embedding_task(document_id: str, content: str, category: str) -> dict:
+    """Generate embedding for a document chunk asynchronously."""
+    import asyncio
+
+    async def _generate() -> dict:
+        from src.infrastructure.ai.openai_client import get_openai_client
+        from src.infrastructure.db import AsyncSessionLocal
+        from src.infrastructure.ai.rag import RAGStore
+
+        async with AsyncSessionLocal() as db:
+            client = await get_openai_client()
+            embedding = await client.create_embedding(content)
+
+            store = RAGStore(db)
+            await store.add_document(
+                content=content,
+                category=category,
+                doc_id=document_id,
+            )
+            return {"document_id": document_id, "embedding_dims": len(embedding)}
+
+    return asyncio.run(_generate())