WhatsApp Copilot
Autonomous WhatsApp CRM & agent supervision workspace
2026 — present
MENA → Global
React 19 • TanStack Router • TanStack Query
Case StudyA native-parity WhatsApp web and desktop workspace equipped with real-time AI reasoning streams, conversational persona mining, autonomous auto-reply controls, and multi-agent monitoring for CRM, sales, and employee supervision.
Ressan's role
Full-stack product engineering and AI architecture
Capabilities
Industry
SaaS & StartupsSystem state
In production · running now
What was built
- 100% pixel-perfect WhatsApp Web and Electron desktop workspace with authentic navigation rail, virtualized chat feeds, and native OS notifications
- Real-time streaming AI thought side panel showing native model reasoning, decision paths, and prompt deliberations
- Relational persona analyzer extracting formality, dialect (Egyptian Arabic, Franco, English), emoji affinity, and relationship dynamics across 1-on-1 and group chats
- Autonomous auto-reply mode with configurable human-in-the-loop safety confirmation modals and banner states
- Supervisor monitoring dashboard tracking agent response times, unreplied chats, and conversational quality across employee accounts
- Marketing broadcast engine with tag-based audience segmentation, custom templates, and campaign delivery tracking
- Multi-provider AI routing abstraction supporting OpenRouter, Gemini, Antigravity local bridge, and WebMCP
- High-throughput WhatsApp REST and MCP gateway backed by PostgreSQL 18 with Prisma ORM and Dokploy deployment
Evidence
No sourced metric on this project.
We shipped this system, but we do not have a measurement we can attribute to it. Rather than estimate a percentage, here is the scope in full — which is a claim you can check with the client.
Delivered
2026 — present
Engagement window recorded for this system.
Surface area
8 modules
Distinct things built, listed on this page.
Stack
13 technologies
Taken from the project's own dependency manifest.
01
The problem
Sales, marketing, and support teams handling high volumes of WhatsApp customer chats face three critical bottlenecks: zero centralized oversight across individual employee phones, the inability to maintain a personalized brand tone at scale, and high latency in turning incoming chat leads into CRM records and scheduled follow-ups.
02
The approach
We built a native-fidelity desktop and web workspace that preserves standard operator muscle memory while embedding an intelligent AI copilot. The engine streams the model's actual step-by-step reasoning in real time, mines relationship personas from chat histories to match dialect and formality, and allows supervisors to monitor unreplied conversations, enforce response SLAs, and trigger automated CRM actions without disrupting direct human connection.
03
The outcome
A comprehensive messaging and CRM workspace that eliminates context switching, reduces operator response times, and gives leadership full visibility into customer communications and automated campaign conversions.
Technology
The stack as shipped.
- React 19
- TanStack Router
- TanStack Query
- Tailwind CSS v4
- shadcn/ui
- Electron
- Node.js
- Prisma
- PostgreSQL 18
- Dokploy
- Gemini API
- OpenRouter
- Docker
Related
