2026.10.04
DeskcommCRM: Bringing WhatsApp Conversations into a Self-Hosted Sales Workflow
DeskcommCRM is an open-source, self-hosted CRM for chat-based sales teams, combining WhatsApp, AI agents, and sales pipelines. Evaluate infrastructure, third-party services, operations, and compliance before adoption.
What the project is for
DeskcommCRM targets teams that acquire, qualify, and close customers through WhatsApp and other chat channels. It combines a CRM, inbox, sales pipeline, and AI agents in a self-hosted application. The project describes the agents as tools for handling conversations, qualifying leads, and supporting sales workflows. GitHub metadata also lists multi-tenancy, MCP, RAG, Supabase, and WAHA among its topics. The repository is presented under the MIT License and is not archived.
The project is a better fit for small and midsize businesses, sales teams, and developers who want control over their deployment and data and are prepared to operate a server. It is less suitable for users looking for a purely managed service with no infrastructure work. Teams already using WhatsApp for customer service may value the ability to connect lead records, pipeline stages, and conversations in one application. However, the project’s positioning as an open alternative does not establish that it replaces commercial products for every scale or industry.
Known features
The README describes WhatsApp connectivity through WAHA and AI integrations for customer interactions. The deployment stack uses Supabase for Postgres, authentication, and storage. Its installation workflow can create the first administrator, apply the database schema, configure HTTPS, and install automation and update agents. The documentation describes both an external Supabase arrangement and a setup with related components on the same server.
Release v1.71.0 added an “open conversation” action to funnel cards and lead records. When a lead has a phone number, the action can reopen an existing conversation or create a missing one; without a number, the action is disabled. The same release added a funnel-analysis API covering stage conversion, time to close, and wins and losses by source. The release notes explicitly say this delivery is an API and that the /app/metrics screen is still a later step.
The release also introduced optional Supabase root-certificate configuration and a database TLS health check. This can help deployments that require stronger certificate-chain verification, but it should not be read as evidence that the project has completed an independent security audit.
Limitations and deployment considerations
This is self-hosted software. Operators need a VPS, a domain, Docker, Supabase settings, an AI provider key, and a WhatsApp connection method. The README recommends 4 GB of RAM, but the supplied sources do not provide a full capacity model, concurrency limit, or performance benchmark. That recommendation therefore cannot establish an appropriate team size. ARM64/aarch64 VPS support is documented, although the selected host, images, and network setup still need to be checked.
AI and WhatsApp are also not supplied entirely by the repository. Operators must choose a provider such as OpenRouter, Anthropic, or OpenAI and connect a number by QR code or use Meta’s official channel. Provider fees, regional availability, messaging policies, and data-processing terms must be reviewed separately; the sources provide no universal pricing or guarantee of third-party availability.
Finally, automatic updates, backups, HTTPS, reverse proxies, and certificates remain production-operations responsibilities. The README describes proxy detection and an update rollback flow, but these do not replace restore testing, access reviews, log management, or privacy assessment. If the system handles personal information, review applicable local regulations and the terms of WhatsApp and the selected AI provider before going live.
Sources
Sources checked: 2026-10-04