How Developers Can Build Their Own WhatsApp CRM + Voice AI Ecosystem Without Vendor Lock-In

The AI communication revolution has quietly entered a new phase.

For years, businesses depended on expensive SaaS platforms for:

• WhatsApp automation
• AI calling agents
• CRM workflows
• Voice bots
• Customer engagement systems

Every month, companies paid:

• WhatsApp API fees
• Voice AI platform charges
• Per-minute calling costs
• LLM usage fees
• Additional platform subscription fees

And despite spending heavily, organizations still did not truly own their infrastructure.

Now that is changing.

Two emerging open-source platforms are reshaping the ecosystem:

1. Open-source WhatsApp CRM & Automation Platform
2. Dograh, an open-source Voice AI orchestration platform

Together, these platforms represent something much bigger:

A fully self-hosted AI communication stack where organizations control:

• their workflows,
• their data,
• their AI models,
• their infrastructure,
• and their long-term costs.

The Problem With Traditional AI Communication Platforms

Most businesses today use platforms like:

• Vapi
• Bland
• Retell
• Twilio workflows
• Proprietary WhatsApp CRMs

These tools are fast to start with.

But over time, companies face several major challenges:

Modern AI systems are no longer simple chatbots.

They are:

• real-time systems,
• event-driven platforms,
• conversational orchestration engines,
• multi-provider integrations,
• API-driven workflow ecosystems.

And businesses increasingly want ownership.

Open Source WhatsApp CRM & AI Voice Revolution

Below is a novice-friendly step-by-step implementation guide combining both platforms:

1. Open-source WhatsApp CRM
2. Dograh Open-source Voice AI Platform

These steps are written so even a fresher can follow and develop a working prototype.

Part 1: Build Open-source WhatsApp CRM

A. Prerequisites

Install these first:

B. Project Setup

Step 1: Clone GitHub Repository

Open terminal:

git clone <github-repo-url>

Example:

git clone https://github.com/your-repo/wacrm.git

Go inside project folder:

cd wacrm

Step 2: Install Dependencies

npm install

This installs all required Node.js packages.

Step 3: Open Project in Antigravity IDE

1. Open Antigravity IDE
2. Click Open Folder
3. Select wacrm folder
4. Click Trust Project

Step 4: Create .env.local

Run:

cp .env.example .env.local

For Windows:

copy .env.example .env.local

Open .env.local.

You will later fill values like:

NEXT_PUBLIC_SUPABASE_URL=
NEXT_PUBLIC_SUPABASE_ANON_KEY=
SUPABASE_SERVICE_ROLE_KEY=
META_APP_SECRET=
ENCRYPTION_KEY=

C. Supabase Setup

Step 5: Create Supabase Project

1. Go to Supabase
2. Login
3. Click New Project
4. Enter:
   • Project name
   • Database password
5. Click Create Project

Step 6: Copy Supabase Keys

Go to:

Project Settings → API

Copy:

NEXT_PUBLIC_SUPABASE_URL=your_project_url
NEXT_PUBLIC_SUPABASE_ANON_KEY=your_anon_key
SUPABASE_SERVICE_ROLE_KEY=your_service_role_key

Paste them in .env.local.

Important: Never share SUPABASE_SERVICE_ROLE_KEY.

Step 7: Run Database Migrations

In project, open:

supabase/migrations/

You will see multiple SQL files.

For each SQL file:

1. Open file in Antigravity
2. Copy full SQL
3. Go to Supabase
4. Open SQL Editor
5. Paste SQL
6. Click Run

Run files in order:

001 → 002 → 003 → ... last file

This creates CRM database tables.

Step 8: Generate Encryption Key

Run:

openssl rand -base64 32

Copy output and paste in .env.local:

ENCRYPTION_KEY=generated_key_here

Step 9: Start CRM Locally

Run:

npm run dev

Open browser:

http://localhost:3000

If port is busy, it may run on:

http://localhost:3003

Step 10: Disable Email Confirmation

In Supabase:

Authentication → Providers → Email

Disable:

Confirm Email

Save changes.

Step 11: Create CRM User

Open CRM URL.

Click:

Create Account

Enter:

• Name
• Email
• Password

Now login to CRM dashboard.

D. Meta WhatsApp API Setup

Step 12: Create Meta Developer App

Go to:

developers.facebook.com

Steps:

1. Login
2. Click Create App
3. Select Connect with customers through WhatsApp
4. Give app name
5. Select business portfolio
6. Click Create App

Step 13: Get WhatsApp Details

Inside Meta app:

WhatsApp → API Setup

Copy:

Step 14: Test Message from Meta

In Meta API Setup:

1. Add recipient phone number
2. Verify OTP
3. Click Send Message

If WhatsApp message comes, Meta setup is working.

E. Create Permanent Access Token

Step 15: Go to Meta Business Settings

Open:

business.facebook.com

Go to:

Business Settings → Users → System Users

Step 16: Create System User

1. Click Add
2. Enter system user name
3. Select type: Admin
4. Click Create System User

Step 17: Assign Assets

Assign:

1. Meta App
2. WhatsApp Account

Give:

Full Control

Step 18: Generate Permanent Token

Click:

Generate Token

Select app.

Set expiry:

Never

Select permissions:

whatsapp_business_management
whatsapp_business_messaging
business_management

Copy token.

F. Configure CRM WhatsApp Settings

In CRM dashboard, go to:

Settings → WhatsApp Config

Enter:

Permanent Access Token
Phone Number ID
WhatsApp Business Account ID
Webhook Verify Token

Example verify token:

my_secure_verify_token_123

Save.

G. Add Meta App Secret

In Meta Developer Portal:

App Settings → Basic

Copy:

App Secret

Paste in .env.local:

META_APP_SECRET=your_app_secret

Restart local server:

npm run dev

H. Webhook Setup Using ngrok

Step 19: Start ngrok

If CRM runs on port 3000:

ngrok http 3000

If CRM runs on port 3003:

ngrok http 3003

You will get HTTPS URL like:

https://abc123.ngrok-free.app

Step 20: Configure Webhook in Meta

Go to:

Meta App → WhatsApp → Configuration

Callback URL:

https://abc123.ngrok-free.app/api/whatsapp/webhook

Verify token:

same_token_used_in_crm

Click:

Verify and Save

I. Create First WhatsApp Automation

Step 21: Create Contact

In CRM:

Contacts → Add Contact

Enter:

• Name
• Mobile number with country code
• Email
• Company

Save.

Step 22: Create Automation

Go to:

Automations → Create Automation

Trigger:

New message received

Condition:

Message contains keyword

Keyword:

Astro

Action:

Send Message

Message:

Hi, thanks for contacting Astro.

Set automation:

Active

Save.

Step 23: Test WhatsApp Automation

From your test WhatsApp number, send:

Hi Astro

Expected reply:

Hi, thanks for contacting Astro.

Now your WhatsApp CRM automation is working.

Part 2: Build Dograh Voice AI Platform

Dograh is an open-source alternative to hosted voice AI platforms like Vapi, Bland, and Retell. It allows developers to self-host voice AI agents with workflows, tracing, recordings, and API calls.

A. Prerequisites

Install:

B. Dograh Setup

Step 1: Clone Dograh Repository

git clone <dograh-github-repo-url>

Go inside folder:

cd dograh

Step 2: Start Dograh with Docker

Run:

docker compose up

or:

docker-compose up

Wait until all containers are running.

Step 3: Open Dograh UI

Open browser:

http://localhost:<dograh-port>

The exact port depends on project configuration.

Common examples:

http://localhost:3000
http://localhost:8080

C. Create Voice AI Agent

Step 4: Create New Agent

Inside Dograh UI:

1. Click Create Agent
2. Give name:

Lead Qualification Agent

3. Select voice provider
4. Select speech-to-text provider
5. Select LLM provider

Step 5: Add Prompt Node

Create first prompt:

You are a polite AI sales assistant. Your job is to qualify inbound leads. Ask the caller what they want to build, their company size, industry, budget, and timeline.

Step 6: Add Qualification Questions

Create flow:

Step 7: Add API Tool Call

Create API tool:

Create or Update Lead in CRM

Example API:

POST https://your-crm.com/api/leads

Payload:

{
  "name": "{{caller_name}}",
  "company": "{{company_name}}",
  "industry": "{{industry}}",
  "budget": "{{budget}}",
  "requirement": "{{requirement}}",
  "call_volume": "{{call_volume}}"
}

Step 8: Add Branch Logic

Create condition:

If budget is high and requirement is clear

Then:

Transfer to human sales team

Else:

Save lead and send follow-up message

Step 9: Add Human Transfer

Transfer condition:

Qualified Lead = Yes

Action:

Transfer Call

Transfer number:

+91XXXXXXXXXX

D. Test Dograh Voice Agent

Step 10: Run Test Call

Inside Dograh:

1. Click Test Call
2. Speak as customer:

We are looking for an AI phone agent for inbound demo requests.

AI should ask follow-up questions.

Step 11: Verify Runtime Logs

After call, check:

This helps debug voice AI properly.

Part 3: Combine WhatsApp CRM + Dograh Voice AI

Now the powerful part.

You can connect both systems into one communication ecosystem.

A. Combined Architecture

Customer
   |
   | WhatsApp Message / Phone Call
   |
-----------------------------------------
|        AI Communication Layer          |
-----------------------------------------
| WhatsApp CRM       | Dograh Voice AI   |
| Meta API           | STT + LLM + TTS   |
| Automation Engine  | Voice Workflow    |
-----------------------------------------
   |
   | API Calls
   |
-----------------------------------------
| CRM / ERP / HMIS / HRMS / Synth Ed     |
-----------------------------------------
   |
   | Database
   |
-----------------------------------------
| Supabase / PostgreSQL / MySQL          |
-----------------------------------------

B. Example Combined Flow

Scenario: Admission Inquiry

1. Parent sends WhatsApp:

I want admission details

2. WhatsApp CRM auto replies:

Thank you. Our AI counselor can call you. Please share preferred time.

3. Parent replies:

Call now

4. CRM calls Dograh API:

Start voice call

5. Dograh AI asks:
   • Student class
   • Location
   • Board preference
   • Budget
   • Transport need
6. Dograh updates CRM lead.
7. Qualified lead is assigned to admission counselor.

C. Example Healthcare Flow

1. Patient sends WhatsApp:

Need appointment with dentist

2. CRM captures mobile number.
3. Dograh voice agent calls patient.
4. AI asks:
   • Symptoms
   • Preferred date
   • Doctor preference
   • Emergency or routine
5. API creates appointment in HMIS.
6. WhatsApp CRM sends confirmation.

Final Learning Outcome

You will learn:

• Git cloning
• Node.js project setup
• Environment variables
• Supabase database setup
• SQL migration execution
• Meta WhatsApp API setup
• Webhook configuration
• ngrok tunneling
• Docker Compose
• Voice AI workflow creation
• API tool calling
• CRM integration
• Basic AI automation architecture