Who DeskcommCRM is for#
E-commerce teams selling on WhatsApp
DeskcommCRM integrates with Nuvemshop and captures leads from landing page webhooks. The AI agent answers product and delivery questions using the catalog you provide, qualifies the lead, and moves them through the funnel. Order and catalog context follows the lead into each conversation without the salesperson needing to look it up.
Skip if:
Your checkout happens on a self-serve web storefront with no chat touchpoints. The value here is in managing conversational sales, not in replacing a web checkout flow.
Service businesses with appointment-based sales
The funnel vocabulary is configurable, so a clinic sets the stage label from 'lead' to 'Patient' and 'won' to 'Scheduled' without forking the codebase. The AI handles initial qualification, the Radar surfaces prospects who expressed interest but never booked, and follow-up automations run on their own schedule.
Skip if:
Your scheduling already has a dedicated booking system with automated reminders. Adding a second follow-up layer for the same contacts may create duplicate outreach.
Agencies setting up CRM instances for clients
Two environment variables swap the name and logo across the entire interface without touching code. The MIT license permits hosting for third parties and charging for it. Each client gets an independent installation with its own knowledge base, funnel configuration, and user roster.
Skip if:
Your clients need custom color schemes or multiple brands on one shared installation. Per-organization branding is not supported; each brand requires a separate VPS deployment.
Teams replacing HubSpot or Freshsales to cut seat costs
At five or fifty salespeople, the cost is the same VPS bill. There is no tier that unlocks features as you grow; the full product is available on install. For teams where CRM seat fees are a recurring budget line item, switching to DeskcommCRM eliminates that cost entirely.
Skip if:
Your sales team depends on deep integrations with email marketing tools, ad platforms, or a data warehouse that expects native CRM connectors. DeskcommCRM's integrations are WhatsApp-focused and still expanding.
The problem it solves#
Sales teams that sell on WhatsApp face a structural challenge: conversations happen in one app, pipeline records live in another, and the two never sync. A lead messages in the morning, someone responds, but by afternoon the CRM has no record of what was discussed or promised. When the follow-up falls through, there is nothing to audit.
The second problem is AI agents that act without guardrails. Most chatbot tools send whatever they generate, which means an agent can quote a price that has changed, promise a delivery window that does not exist, or contact someone who has opted out. When the customer complains, there is no record to review: no trace of what the agent considered, what it nearly sent, or why it chose those words.
How it solves it#
Seven-layer AI message guardrails
Before each AI message is sent, the system runs seven checks in fixed order: unsubscribe status, LGPD compliance, anti-ban throttling, text variation, deterministic promise checking, semantic promise checking, and automation disclosure. Each check is logged as a durable record, including messages the agent decided not to send and the reason it blocked them.
RAG knowledge base per tenant
Each organization gets its own knowledge base that the AI reads before responding. The agent answers questions from your catalog, policies, and delivery windows instead of generating responses from general training data. You update the knowledge base from the interface; changed content is indexed for the next conversation without touching code.
WHEN/IF/THEN automation rules
Any CRM event (lead stage change, new tag, incoming WhatsApp message, webhook trigger) can fire actions: add a tag, move the lead, assign to an agent, send a WhatsApp message, or call an external webhook. Rules are created in the UI, always start paused for review, and run from a queue that processes every minute.
Lead Radar for stalled follow-ups
Open leads without a scheduled next step appear in the Radar screen, classified by time without interaction and what is pending. Stalled deals are visible in real time, not in a next-month report. Each entry shows current risk level so a salesperson can prioritize without reviewing every open deal manually.
Multi-tenant with RLS isolation tested in CI
Multiple organizations share one installation but are fully isolated at the database level. Row-level security on every tenant-aware table is validated by 364 invariant tests in CI, including a test that creates two organizations and proves neither can see the other's rows, plus a control case that verifies the rows exist before the isolation check runs.
WhatsApp via QR code or Meta Cloud API
Connect via WAHA for immediate QR code setup with multiple numbers and anti-ban protections (throttle, jitter, scheduled hours). Or use Meta's Cloud API for an official WhatsApp channel with approved message templates. Both modes sync media to private Supabase Storage with signed URLs and support STOP detection.
Strengths and trade-offs#
Strengths
- MIT license with no per-user pricingThe full product is available under MIT with no paid tier and no features locked behind a subscription. Unlike Freshsales or HubSpot Sales Hub, which charge per seat each month, the cost here is a fixed VPS bill regardless of how many salespeople you add. Five people or fifty cost the same server.
- Auditable AI with durable decision logsEvery AI decision is stored: what the agent considered sending, which guardrail blocked it, and why. This includes messages the agent drafted but did not send. The audit trail is durable and exportable, so you can review any message event from the conversation timeline rather than reconstructing it from customer complaints.
- One-command install with a tested update pathThe setup script handles Docker, Postgres schema, HTTPS certificate, and cron configuration in one pass. Updates run from a UI button or a single terminal command, with automatic database backup before anything changes. The update path is tested in CI against both fresh and existing databases to verify idempotency.
- LGPD compliance built into the message pipelineBrazilian data protection law compliance is built into the AI guardrails (checked on every outbound message), the audit log (append-only, five-year retention), and data handling (anonymization preferred over deletion). If you host on a VPS with a Brazilian datacenter, personal data does not cross international borders.
Trade-offs
- -Requires a VPS with at least 4 GB RAMThe stack runs 7 Docker containers, and each active WhatsApp session consumes approximately 150 MB. The documentation states a 2 GB VPS can start the stack but operates at its limit. Shared hosting does not work; a dedicated VPS is required. The 4 GB recommendation gives headroom for multiple simultaneous WhatsApp sessions.
- -Documentation and installer are Portuguese-firstThe product website, README, and installer prompts are in Brazilian Portuguese. An English README exists in the repository but the installer interface and setup guides are Portuguese-first. Teams without Portuguese readers will need to translate setup instructions or use the Claude Code-based installation assistant described in the README.
- -Young project with 89 open issuesThe repository was created in April 2026 and had 89 open issues at the time of this review. The project is actively maintained (last commit on the same day as this review) but is under a year old. Some edge cases and integrations are still being worked through; check the issue tracker before relying on a specific workflow or integration.
DeskcommCRM vs alternatives#
DeskcommCRM vs HubSpot Sales Hub
HubSpot Sales Hub and DeskcommCRM both track leads through a pipeline and automate outreach. The key difference is deployment model: HubSpot is a cloud-hosted service with per-seat subscription pricing; DeskcommCRM runs on your own VPS with a flat server cost.
| Feature | DeskcommCRM | HubSpot Sales Hub |
|---|---|---|
| License | MIT (self-hosted) | Proprietary |
| Self-hosting | Yes | No |
| WhatsApp native | Yes | Via third-party add-on |
| Pricing model | VPS cost only | Per-seat subscription |
| AI agents | Built-in with audit logs | AI features on higher tiers |
| Data location | Your infrastructure | HubSpot's cloud |
DeskcommCRM is the stronger choice when your primary sales channel is WhatsApp, your team operates under LGPD, or you need to eliminate per-seat cost as headcount grows. HubSpot is still the better choice when you need its native ecosystem (email marketing, ad attribution), a polished mobile CRM experience, or enterprise-grade support with SLAs. Its email sequencing and marketing automation depth are not replicated in DeskcommCRM.
DeskcommCRM vs Freshsales
Freshsales targets teams that want affordable CRM with built-in phone, email, and a clean pipeline UI. DeskcommCRM targets teams whose primary channel is WhatsApp and who want self-hosted infrastructure.
| Feature | DeskcommCRM | Freshsales |
|---|---|---|
| License | MIT (self-hosted) | Proprietary |
| Self-hosting | Yes | No |
| WhatsApp native | Yes | Requires Freshchat add-on |
| Pricing model | VPS cost only | Per-seat subscription |
| AI compliance | LGPD guardrails built-in | Configurable, not built-in |
| Audit trail | Every AI decision logged | Limited AI audit visibility |
Freshsales is the better fit when you need built-in telephony, native email campaigns, a mobile app, or a team already in the Freshworks ecosystem. DeskcommCRM wins on WhatsApp-native operations, transparent AI behavior with full audit logs, LGPD compliance by default, and zero per-user cost.
DeskcommCRM vs Salesforce Sales Cloud
Salesforce Sales Cloud is the enterprise CRM benchmark with deep customization, a large integration ecosystem, and support for complex multi-stage sales workflows. DeskcommCRM is not a replacement for mid-market or enterprise Salesforce deployments.
| Feature | DeskcommCRM | Salesforce Sales Cloud |
|---|---|---|
| License | MIT (self-hosted) | Proprietary |
| Self-hosting | Yes | No |
| WhatsApp native | Yes | Via AppExchange partners |
| Pricing model | VPS cost only | Per-seat subscription |
| Setup | One bash command | Requires dedicated admin |
DeskcommCRM is a practical alternative to Salesforce only for small teams where the primary channel is WhatsApp and the seat cost is the blocking constraint. Salesforce's reporting depth, territory management, and integration ecosystem are not replicated here. If your organization relies on a Salesforce admin and its AppExchange partners, DeskcommCRM does not cover that scope.
Quick start#
Self-hosting deploys the full stack on any VPS; the install script handles Docker, HTTPS, and database setup automatically.
```bash
git clone https://github.com/melgarafael/DeskcommCRM.git
cd DeskcommCRM
bash hostgator-setup-kit/install.sh
```What it's built on#
- Languages
- TypeScript
- Frameworks
- Next.jsReact
- Databases
- PostgreSQL
FAQ#
Does DeskcommCRM cost anything to run?
The software is free and MIT-licensed. You pay only for the VPS it runs on (4 GB RAM recommended) and any AI API keys you use for the agents (OpenRouter, Anthropic, or OpenAI). There is no paid tier, no per-user fee, and no features locked behind a subscription. The full product is available on install.
What WhatsApp connection options are available?
Two options: WAHA via QR code scan for quick setup with multiple numbers and anti-ban protections (throttle, jitter, scheduled sending hours), or Meta's Cloud API for an official WhatsApp channel with approved message templates. You choose at install time and can switch later from the Connections screen in the admin interface.
Is DeskcommCRM compliant with Brazilian data protection law (LGPD)?
LGPD compliance is built in. The AI message guardrails check consent and unsubscribe status before every automated outbound message. The audit log is append-only with five-year retention. Data anonymization is the preferred approach over deletion. If you host on a VPS with a Brazilian datacenter, personal data does not cross international borders under the ANPD Resolution 19/2024 standard.
How do updates work after the initial install?
Either click the 'Nova versão' button in the sidebar (which triggers an automatic update with database backup), or run bash update.sh from the terminal. Both paths back up the database first, apply the schema idempotently, pull the new image, and verify service health. The update path is tested in CI, not just documented.
Can I install DeskcommCRM for multiple clients and charge for it?
Yes. The MIT license permits commercial use, modification, and redistribution, including hosting for third parties. Two environment variables in .env swap the brand name and logo across the interface. Each client gets a separate installation. Per-organization branding on a shared installation is not supported.
Similar open-source tools#
OpenFang
Open source Agent OS built in Rust with autonomous agents
Twenty
Open source CRM with custom data models and AI-native workflows
EspoCRM
Self-hosted open source CRM for contacts and pipelines
FckSignups
Open-source tools that work instantly, no signup required
fmt
Fast, type-safe C++ formatting that replaces printf and iostreams
dograh
Open source voice agent builder with telephony and BYOK AI support