DeskcommCRM

Product Introduction

1. Definition: DeskcommCRM is a self-hosted, open-source commercial operating system (CRM) built with a containerized architecture (Docker). It functions as a live sales pipeline management platform that integrates native AI agents for WhatsApp communication.
2. Core Value Proposition: It exists to eliminate lost leads in the sales funnel by replacing inefficient spreadsheets and unaccountable chatbots. It provides businesses that sell via chat with complete control, auditability, and no per-user monthly fees, ensuring no lead dies without a diagnosis.

Main Features

1. Native WhatsApp AI Agents: AI agents operate directly within the user's WhatsApp account. They read conversation history and company knowledge base before responding, qualify leads based on configurable funnel stages, and log all actions. The technology stack uses containerized services to manage the WhatsApp Web session and AI processing securely on the user's own server.
2. Seven-Step Guardrail System: Every AI-generated message undergoes seven deterministic, versioned checks before sending: opt-out (descadastro), LGPD compliance, anti-ban (platform rules), text variation, deterministic promise check, semantic promise check, and automation disclosure. Each check, including those that block a message, creates a durable, exportable audit log (lib/agent-engine/guardrails/before-send.ts).
3. Live System Doctrine & Radar: The platform is governed by a public "live system" doctrine (docs/doctrine/sistema-vivo.md). A core feature is the "Radar," a real-time dashboard that classifies all open demand by risk (critical, at risk, in flight) based on time without interaction and pending tasks, preventing leads from going stale unnoticed.
4. Self-Hosted, One-Command Installation: Deployment is designed for a VPS via a single Bash script (install.sh). The script handles dependencies, secret generation, database schema application, Docker container orchestration (7 containers), and first admin creation. It features automated HTTPS certificate issuance and a tested, non-breaking update path (update.sh).
5. Multi-Tenant Ready & Configurable Funnel: The system supports multi-niche operations out-of-the-box. Funnel stage vocabulary (e.g., Lead→Payment→Won for e-commerce; Patient→Scheduled→Won for clinics) is configurable via environment, not code forks. Row-Level Security (RLS) ensures strict data isolation between organizations, verified by 364 invariant CI tests (tests/invariants/rls-isolation.test.ts).

Problems Solved

1. Pain Point: The "pretty spreadsheet" CRM problem, where leads are logged but then lost because the system isn't responsible for what happens next. There is no visibility into why a lead went cold or who was accountable.
2. Pain Point: The "robot that replies and vanishes" problem, where chatbots make unverified promises (like non-existent delivery times) with no audit trail, creating business risk and customer distrust.
3. Target Audience: Brazilian SMBs and agencies selling via conversational channels (WhatsApp), including e-commerce, clinics, real estate, info-products, and service providers who need full data sovereignty and cost control.
4. Use Cases: A clinic automating patient scheduling and follow-up via WhatsApp; an e-commerce store qualifying leads and answering product/delivery FAQs 24/7; a marketing agency installing and managing white-labeled instances for multiple clients under the MIT license.

Unique Advantages

1. Differentiation: Unlike SaaS CRMs (like RD Station, PipeDrive) that charge per user/month and hold data on their cloud, DeskcommCRM is a one-time infrastructure cost (VPS) with full data ownership. Unlike other open-source CRMs, it has native, auditable AI agent integration for WhatsApp built into its core doctrine.
2. Key Innovation: The "Live System" doctrine, which is a public, codified set of operational rules (like "No lead dies without diagnosis") that every feature must satisfy. This transforms the software from a passive database into an active, accountable commercial operating system. The seven pre-send guardrails for AI provide a unique layer of deterministic safety and compliance.

Frequently Asked Questions (FAQ)

1. How much does DeskcommCRM cost? The software is completely free under the MIT license, with no paid version or feature locks. The only costs are the VPS hosting (e.g., HostGator's NVMe 4 plan) and the consumption of AI API keys (OpenAI/Anthropic).
2. Do I need to know how to code to install DeskcommCRM? No programming is required, but you need to follow the technical installation guide and have server (VPS) access. The process is command-line based (bash install.sh), and an AI-powered assistant (Claude Code) is included to guide the setup conversationally.
3. Is my data kept in Brazil with DeskcommCRM? Yes, if you use a VPS with a datacenter in Brazil. Since the system is self-hosted, all data (conversations, leads, logs) resides on your server, avoiding international data transfer and complying with ANPD Resolution No. 19/2024.
4. Can I use DeskcommCRM for my agency's clients? Yes. The MIT license explicitly permits commercial use, modification, and redistribution. You can host separate instances for clients, and basic white-labeling (name/logo) is possible via environment variables without code modification.
5. What happens if the AI makes a mistake? Every AI action is auditable. The seven guardrails are designed to catch errors (e.g., promising a non-existent delivery time). Any blocked message is logged with the reason. For complex scenarios, the agent hands off to a human with a full context summary of what transpired.