WHATS
CRM
HUB
About the Technology
Modern Stack
Built on Technology Designed for Scale
It is not just about features โ the engine behind WhatsCRM Hub is designed to broadcast millions of messages, serve true multi-tenant SaaS, and run 24/7 without drama.
The latest version of WhatsCRM Hub is fully rewritten using Go (Golang) on the backend and Vue 3 + TypeScript on the frontend. This is not a trend-chasing decision โ it is the result of hands-on experience serving thousands of users across Indonesia, Malaysia, Saudi Arabia, and Brazil.
The result: a lightweight deployment (single binary under 20MB), low RAM usage, low latency even during massive broadcasts, and true multi-tenant isolation between customers.
Backend
Backend: Go (Golang)
The language of choice for world-class infrastructure teams (Google, Cloudflare, Uber, Twitch). Built for concurrency and simple deployment.
Go 1.22+ & Fiber v2
Web framework powered by fasthttp. Capable of handling tens of thousands of requests per second per instance with a tiny RAM footprint.
PostgreSQL 15+ & sqlc
The most mature relational database for multi-tenant SaaS. Type-safe queries โ no N+1, no SQL injection, ORM-like ergonomics without ORM overhead.
Asynq Queue (Redis-backed)
A worker pool with mature retry, dedup, and visibility. Every broadcast, follow-up, and webhook flows through a distributed queue you can monitor in real time.
Native WebSocket (coder/websocket)
Context-aware realtime chat โ connections clean up automatically when tabs close. No Socket.IO bloat.
JWT + Refresh Rotation
In-memory access token, httpOnly cookie refresh, CSRF protection, multi-tab coordination via BroadcastChannel. Aligned with current OWASP best practices.
zerolog + Structured Logging
Every request has a request-id. Logs ready to pipe into Loki / Datadog / CloudWatch with no extra parsing.
Frontend
Frontend: Vue 3 + TypeScript
A modern, reactive, type-safe SPA. The ecosystem chosen by GitLab, Alibaba, and hundreds of productive SaaS teams.
Vue 3 Composition API
The most efficient reactivity engine in its class. Small components, easy to test, easy to maintain.
End-to-End TypeScript
Bugs caught in your IDE, not in production. Every API response has the same type as the Go struct on the backend.
Vite + Pinia
The fastest build tool (hundreds of times faster than Webpack), with lightweight, explicit state management. Sub-second hot reload.
PrimeVue v4 UI
Enterprise-grade UI components โ virtual-scroll DataTable, timezone-aware Calendar, async AutoComplete. Consistent across devices and browsers.
Broadcast Engine
A Broadcast Engine That Holds Up Under Volume
Send hundreds of thousands of messages a day without crushing your server, without disturbing other tenants, without hitting Meta rate limits.
50 Concurrent Worker Pool
50 goroutines working in parallel โ compare that to PHP-FPM that needs one process per request. Hundreds of thousands of messages eaten in minutes, not hours.
Token Bucket: 250 RPS / Phone
Per-sender rate limit โ automatic protection from Meta Graph API blocks. You just send; the system handles pacing.
Anti Soft-Ban (Jitter 2โ5s)
For WhatsApp Web โ random spacing between messages so your sending pattern looks natural, reducing the risk of WhatsApp banning your number.
50,000 Queue Backpressure
When the queue is full, the system rejects new jobs gracefully โ no crash, no OOM. Customers get a clear error message, not mysterious timeouts.
Multi-Channel Native
WhatsApp Web (whatsmeow), WhatsApp Business API (official Meta Graph), Instagram DM, Messenger โ one inbox, one worker pool, one bill.
Automatic Exponential Retry
A message fails? The system retries with smart backoff. You only see messages that truly cannot be delivered, not transient glitches.
Business Impact
What This Means for Your Business
Every technology decision we made comes back to one thing โ so you can focus on selling, not fixing servers.
6GB
RAM Enough for 100k Messages/Day
A cheap VPS can handle it. No need for enterprise-class servers for SMB to mid-market scale.
< 20MB
Deployable Binary Size
One executable file. No 800MB node_modules, no PHP-FPM tuning, no dependency hell when installing on a new VPS.
< 100ms
API P99 Latency
Fast response = happy users = better retention. Dashboards, chat, and visitor widgets feel instant.
โ
Isolated Multi-Tenant
Tenant A broadcasting millions of messages will never disturb Tenant B. Perfect for SaaS and reseller models.
Tech Comparison
Why Not Node.js + Baileys?
A question we get often from developers and prospective partners. Here is the honest answer, from hands-on experience building this product.
Go + whatsmeow (our choice)
- Native concurrency โ cheap goroutines can hold thousands of WhatsApp Web connections in a single process without lag.
- Single binary deploy โ copy the file, run it, done. Perfect for self-hosted customers without a DevOps team.
- Stable RAM โ Go has sub-millisecond garbage collection. No 200ms freezes when many WhatsApp sessions are connected (a classic Node.js problem).
- Mature Asynq queue โ retries, dedup, visibility, scheduler โ all built-in. No need to wire it together yourself like BullMQ.
- End-to-end type safety โ pattern matching bugs, race conditions, nil pointers โ caught by the compiler before production.
- whatsmeow โ a WhatsApp Web library used by many serious projects, more stable, with documented protocol (Signal protocol).
Node.js + Baileys (alternative)
- Single-thread event loop โ only one CPU core per process. For large broadcasts you must run many instances, adding complexity.
- Heavy node_modules โ hundreds of MB of dependencies, frequent breaking changes, weekly security audits.
- Long GC pauses โ when WhatsApp sessions pile up, the V8 garbage collector can freeze for hundreds of milliseconds. Users say "this feels slow."
- Baileys is fragile at scale โ great for hobby and prototypes. Production multi-tenant with hundreds of devices often hits memory leaks and reconnect storms.
- PM2 + Redis + Nginx + Node โ a stack that must be tuned piece by piece. Self-hosted customers often give up at setup.
- TypeScript optional โ many libraries still ship plain JS, type safety is inconsistent across the dependency tree.
In short: Node.js + Baileys is great for experiments, Go + whatsmeow is built to be sold to paying customers. That is the difference โ and that is what we picked.
Curious to See This Engine in Action?
Book a 30-minute demo. We will walk you through the dashboard, realtime broadcasts, and the backend architecture running behind the scenes.