Hey Product Hunt 👋 Kamal here — founder of MyAgentMail. I built this because every outbound or research agent I tried to ship hit the same two walls: Agents can't receive email. Sure, you can hand them a Gmail OAuth token and they'll send — until the token expires mid-cadence, gets revoked because the human read a suspicious-login email, or never exists on day one of a SaaS product. And even when sending works, inbound replies route nowhere an agent can see. Agents can't sit and watch for intent. LinkedIn scrapers tell you what's there right now. They can't sit in the background for two weeks and fire a webhook the moment a founder complains about cold email or a hiring manager posts "looking for a head of growth." MyAgentMail solves both as primitives, not products. On the email side: real two-way threaded inboxes on custom domains, with proper Message-ID / In-Reply-To so replies land on the same thread automatically. SMTP/IMAP credentials per inbox if a human needs to jump in. Inbound webhooks for message.received, message.bounced, message.complained. Attachments. Drafts. The mail-client verbs you'd expect, exposed as a JSON API. On the LinkedIn side: plain-English firing rules ("flag founders complaining about cold email"), continuous polling, webhook-on-match. Daily quota grows with each LinkedIn account you connect — not with how much you pay us. Plus full programmatic access to connect, DM, post, search, list connections, with inline-2FA-over-WebSocket so security challenges don't break automation. Stitch them together with cadences (linkedin_connect → wait 2d → DM → wait → email fallback) where each message body can come from your own webhook in your own voice, with reply detection across both channels stopping the cadence the instant a human responds. Who it's for: builders shipping outbound, research, or recruiting agents. Multi-tenant SaaS that needs to give every customer isolated email + LinkedIn primitives (workspace-scoped wk_ keys handle isolation). Anyone who's tried to ship an "email agent" and hit OAuth hell. Pricing. Email from $5/mo, LinkedIn from $29/mo, mix-and-match. Cadences + webhooks included on every tier. There's a TypeScript SDK (npm install myagentmail), a public OpenAPI spec, and MCP servers for every primitive so Claude / Cursor / your agent runtime can call them directly. What I'd genuinely love feedback on: For builders: what agent did you try to ship that hit the "can't receive email" wall? I want to know if the use cases I'm betting on are real. For non-builders: does "real inboxes for AI agents" land, or does it sound like another email tool? The positioning is the thing I'm least sure about. Happy to go deeper on any specific feature — threading, deliverability infra, cadence runner, intent-signal LLM matching, workspace isolation. I read every reply. – Kamal

MyAgentMail

Outreach infrastructure for AI agents - Email + Linkedin

Outreach infrastructure for AI agents - Email + Linkedin