Mails.ai

About Mails.ai

Mails.ai gives an AI agent its own real email address in a single API call — send, receive and reply — and is built around the assumption that the agent, not a human, is reading the inbox. Every inbound message arrives as a structured reply event carrying an injection score and a sender reputation figure, with intent and extracted entities available as an opt-in classification step, so an agent acts on typed fields rather than parsing a message body. A six-category prompt-injection scanner runs on every inbound and flags the event as quarantined above a 0.95 score, which is the security posture the whole product is organised around: an agent with an inbox is an agent with an untrusted input channel, and the platform treats it that way by default. Cold email is blocked at the API rather than in a terms-of-service clause, which is a deliberate deliverability decision — the reputation graph only works if the platform is not also carrying spam. Free sends come from a workspace subdomain with zero DNS setup; paid plans add custom domains verified with DNS records and per-domain DKIM signing. It ships a TypeScript SDK, REST API, HMAC-signed webhooks with an SDK reactor, per-agent allowlists and budgets, and an MCP server. The vendor states plainly that it has no sales team and no contact-us enterprise tier.

Key Features

Tags

More AI Agent Infrastructure tools