Deploy and Host agente.dev on Railway

agente.dev is a self-hosted builder for AI customer service agents on official WhatsApp. A guided setup turns your answers into a system prompt, and a dashboard lets you test the agent, edit it in plain language, add a knowledge base, and publish it. The interface is in Brazilian Portuguese.

About Hosting agente.dev

Hosting agente.dev takes two services: the app, which runs from a prebuilt Docker image, and PostgreSQL with pgvector, which stores conversations and knowledge base embeddings on a persistent volume. The template connects them, generates the secrets, and creates a public domain. After deploying, open the domain, enter the setup token from the app's variables, and create your login. The onboarding asks for your OpenAI API key and builds the agent. Then connect a WhatsApp number in Integrações by pasting the generated webhook URL into your provider, and click Publicar. Database migrations and the background job queue start with the app, so there is no separate worker.

Common Use Cases

Answer customer questions on WhatsApp using your FAQ, price lists, and policy documents
Qualify leads on WhatsApp and hand the conversation to a person when it is ready
Reply to voice messages, which are transcribed automatically
Send contact, message, and handoff events to a CRM or an n8n workflow

Dependencies for agente.dev Hosting

An OpenAI API key. Usage is billed to your own OpenAI account.
A WhatsApp number connected through dispara.dev or the WhatsApp Cloud API.
PostgreSQL with pgvector, included in this template.

Implementation Details

The template deploys these services:

Service	Image	Notes
`app`	`andreprezoto/agente-dev:latest`	Public HTTP on port `8080`, healthcheck at `/api/health`
`pgvector`	`pgvector/pgvector:pg17`	Volume mounted at `/var/lib/postgresql/data`

The app service uses these variables, all filled in by the template:

Variable	Purpose
`DATABASE_URL`	Connection to the `pgvector` service
`MASTER_KEY`	Encrypts the OpenAI key and WhatsApp tokens. Do not change it after first use, or the saved secrets can no longer be read.
`SETUP_TOKEN`	Protects the first access, so only you can create the login
`PORT`	Port the app listens on

Default OpenAI models, editable in the dashboard's settings:

Use	Model
Text replies	`gpt-6-luna`
Audio transcription	`whisper-1`
Knowledge base embeddings	`text-embedding-3-small`

Each installation runs one agent on one WhatsApp number.

Why Deploy agente.dev on Railway?

Railway is a singular platform to deploy your infrastructure stack. Railway will host your infrastructure so you don't have to deal with configuration, while allowing you to vertically and horizontally scale it.

By deploying agente.dev on Railway, you are one step closer to supporting a complete full-stack application with minimal burden. Host your servers, databases, AI agents, and more on Railway.