Introduction

Portarus sits between your message source and your AI agent. Point your source at a Portarus gateway URL instead of your agent's endpoint, and every message is filtered, deduplicated and delivered with a full audit trail — one URL, zero code changes.

Who it's for

Anyone running an AI agent that receives messages from the outside world — WhatsApp, Telegram, Twilio, Slack or a custom webhook — and wants to keep spam, duplicates and token-stuffed payloads away from the model without writing filtering code.

What you get

• Mechanical filters — rate limiting, deduplication and a token budget, on by default.
• Message unification — rapid-fire messages batched into one coherent payload.
• Keyword routing — send messages to different endpoints based on their content.
• A delivery envelope — every payload arrives with a _portarus object telling your agent exactly what happened.
• Insights — live traffic, costs saved, a message-flow funnel, quarantine and logs.

The shape of it

A gateway is a single URL of the form gateway.portarus.com/g/{slug}. You create one per source/agent pair, configure its filters, and swap it in for your current webhook. The gateway answers your source immediately and delivers the cleaned message to your agent in the background.