AgentSDR icon

AgentSDR

AgentSDR is an open-source, self-hosted sales automation platform for founders and sales teams running email sequences, LinkedIn campaigns, and WhatsApp outreach. It combines a shared lead database with AI reply classification and human-reviewed response drafts, using the organization’s own connected accounts and model key.

AgentSDR

Overview

AgentSDR is an open-source, self-hosted sales development platform for coordinating outbound email, LinkedIn, and WhatsApp activity from one workspace. It combines channel-specific campaign tools with a shared lead database and an AI CRM that classifies replies and prepares response drafts.

The product is designed for teams that want to connect their own mailboxes, social accounts, model provider, storage, and enrichment services. AI drafts are grounded in a knowledge base and held for approval, while channel guardrails such as sending caps, working windows, warm-up rules, and do-not-contact handling are applied across outreach workflows.

Core features

Email sequences

Build multi-step email campaigns from Google Workspace mailboxes using CSV or XLSX columns as merge fields, spin text, per-mailbox caps, sending windows, unsubscribe handling, and bounce suppression.

LinkedIn campaigns

Run invitation and follow-up campaigns across multiple LinkedIn accounts through Unipile, with randomized pacing and stated account limits of 30 premium invites or 5 free invites per day.

WhatsApp calls and messages

Place calls inside WhatsApp Web through a Chrome extension, record and transcribe both sides to the connected R2 bucket, and synchronize messages.

Shared lead database

Import and share people and company records across channels, add custom columns, match duplicates by email or LinkedIn, and create enrichment or campaign workflows from tables.

AI CRM and inbox

Classify replies into configurable stages, draft responses from a knowledge base, queue follow-ups for review, and provide a keyboard-oriented inbox for triage.

Self-hosted data and model control

Run AI calls on a selected provider through the organization's OpenRouter key, with data and recordings stored in the organization's PostgreSQL database and storage bucket.

Use cases

  • Founder-led outbound

    Founders can coordinate founder-led outbound across email, LinkedIn, and WhatsApp while keeping replies in one inbox and approving suggested responses before they are sent.

  • Multi-channel prospecting

    Sales teams can segment imported CSV or XLSX leads, add custom fields and enrichment columns, then turn those records into channel-specific campaigns with shared contact history.

  • Account and mailbox operations

    Teams running several mailboxes or LinkedIn accounts can distribute activity across connected accounts and monitor sending caps, windows, replies, bounces, and unsubscribes in one workspace.

  • Reply triage and response preparation

    Organizations with internal knowledge bases can use AI to classify incoming replies and prepare answers grounded in pricing, FAQs, case studies, and objections, leaving uncertain cases for human review.

  • Self-managed sales infrastructure

    Privacy-conscious or technically capable teams can run the application on infrastructure they control, connect their own provider accounts, and keep leads, conversations, and recordings in their own systems.

Pros and Cons

Pros

  • MIT-licensed code can be used, modified, self-hosted, distributed, or sold, including as a hosted service.
  • One workspace combines email, LinkedIn, WhatsApp, shared lead data, reply triage, and campaign measurement.
  • Organizations use their own provider credentials and model key, with encrypted integration credentials and data stored in their own database and bucket.
  • Human approval remains in the loop for AI-generated replies, with low-confidence classifications waiting for review.

Cons

  • Self-hosting requires operating Docker/PostgreSQL infrastructure and configuring application secrets.
  • The product depends on separately connected third-party services for channels, AI, storage, and enrichment; those providers bill directly.
  • Published pricing and commercial support terms are not available on the supplied pricing page.

FAQ

How is AgentSDR deployed?

AgentSDR can be self-hosted with Docker Compose. The documented setup uses one Next.js application, PostgreSQL, a setup service, and a cron sidecar; the app is available locally on port 3000 after startup.

Which services can AgentSDR connect to?

The product connects Google Workspace for email, Unipile for LinkedIn and WhatsApp, OpenRouter for AI calls, Cloudflare R2 for call recordings, and enrichment providers such as Apollo, Hunter, and Lusha. Connections are configured per organization from the relevant settings pages.

How are integrations and credentials handled?

Each organization supplies its own provider credentials. Credentials are checked with a live provider call before saving and stored encrypted with AES-256-GCM. Features remain unavailable until their required integration is connected.

Does AgentSDR automatically send AI-generated replies?

The AI classifies replies into Interested, Customer, Not interested, or Other, and drafts responses from the organization's knowledge base. Drafts are held for human approval, while low-confidence classifications wait for review.

How much does AgentSDR cost?

The site does not publish pricing or plan details. The pricing URL currently returns a not-found page, so costs for AgentSDR and connected providers should be confirmed separately.

Quick Facts

Category
Sales automation / AI CRM
Deployment
Self-hosted; Docker Compose or source installation
License
MIT
Channels
Email, LinkedIn, and WhatsApp
Database
PostgreSQL 16 or newer
AI setup
Bring your own OpenRouter/model provider key
AgentSDR - AI Tool, Features, Use Cases & Alternatives | UStack