AgentSDR

AgentSDR is an open-source, MIT-licensed outbound workspace led by Kandid. It runs email sequences, LinkedIn campaigns and WhatsApp messages and calls, with an AI CRM that reads every reply and drafts the answer. It is self-hosted on your own server with your own AI model key; the latest GitHub release listed is v0.3.1.

What It Is

AgentSDR is a self-hosted sales engagement platform that replaces a stack of separate lead database, enrichment, sequencing, LinkedIn, dialer and CRM tools with one Next.js app on one PostgreSQL database. Leads are imported from CSV or XLSX, enriched in tables, contacted across channels, and tracked in a shared pipeline with analytics by channel.

How the Outreach Workflow Runs

Email sequences send from your own Google Workspace mailboxes with merge fields, spin text, per-mailbox daily limits (30 by default), sending windows, one-click unsubscribe and bounce suppression. LinkedIn campaigns run through Unipile with an invite, an accept message and follow-ups, paced at 30 invites a day on premium and 5 on free accounts. WhatsApp supports message campaigns and calls dialed inside WhatsApp Web by a Chrome extension, recorded to your Cloudflare R2 bucket and transcribed by your chosen model.

AI CRM and Data Ownership

Every reply on every channel lands in one inbox, is classified (for example interested, customer, not interested, other) and gets a draft grounded in your knowledge base. Nothing sends on its own: drafts wait for a person to approve or edit, and the AI only moves a lead forward when confident. Enrichment tables combine AI columns, HTTP calls, formulas and 15 enrichment providers such as Apollo, Hunter and Lusha. Leads and conversations live in your Postgres, and integration credentials are encrypted with AES-256-GCM.

Setup Path

Deployment needs a machine with Docker and PostgreSQL 16 or newer; the Compose file brings its own database. Download docker-compose.yml, set secrets, run docker compose up -d, then create the admin account and connect services in Settings. It can also be run from source with Bun 1.2+. One deployment can hold several organizations with owner, admin and member roles. The project's responsible-use guidance notes that LinkedIn automation, bulk WhatsApp messaging, cold email and call recording carry legal and platform risks.