deskcommcrm is a CRM application with native AI agents that answer, qualify, and sell to customers on WhatsApp through per-tenant retrieval-augmented generation. The project provides an open-source AI sales operating system positioned as an alternative to Kommo, Octadesk, and Intercom.
Project overview
It combines a CRM architecture with AI agents acting as first-class pipeline assignees and a configurable pipeline vocabulary that spans multiple niches including Customer, Patient, and Buyer.
Project type
AI Agent · RAG
Use cases
Knowledge Q&A
Deployment
Refer to project documentation
License
MIT
Best for
Teams that need a CRM with WhatsApp integration and want AI agents to handle lead answering and qualification through per-tenant RAG.
Organizations operating across multiple niches that require configurable pipeline vocabulary such as Customer, Patient, or Buyer.
Key capabilities
AI agents with per-tenant RAG that answer, qualify leads, and move them through the pipeline, acting as first-class assignees.
Capture sources for incoming leads and define WHEN/IF/THEN rules to automate tagging, lead movement, and notifications.
Limitations and risks
Setup requires configuring multiple third-party services including Supabase, WAHA, Upstash Redis, the Vercel AI SDK via AI Gateway, and Sentry, taking approximately 60 to 90 minutes.
The project is provided on an 'as-is' basis with community-based support and no service-level agreement. Adopters should evaluate whether this support model meets their operational requirements.
Getting started
Clone the repository, install Node 22 and pnpm, copy .env.example to .env.local, optionally run a local WAHA instance via docker compose for development, apply Supabase migrations, then start the application with pnpm dev. Configuring multiple third-party services takes approximately 60 to 90 minutes.