DEV Community: David

Why Most WhatsApp Bots Fail — And How I Solved It with Conversation Memory

David — Thu, 14 May 2026 11:12:48 +0000

The Problem

Here's what happens in the real world:

Monday 10am:
"How much does your service cost?"

Bot: "Our service costs $X. Learn more at..."

Friday 3pm:
"Actually, do you have a payment plan?"

Bot: "Our service costs $X. Learn more at..." (same generic response)

The customer sees:

A bot that doesn't listen
A bot that doesn't remember
A bot that isn't intelligent

They leave. Your competitor wins.

This wasn't a technical problem – it was a business problem. I was losing leads because my bot couldn't maintain a conversation.

The Breakthrough: Conversation Memory

The solution clicked when I thought about how humans sell: we remember everything about our customers.

What if the WhatsApp bot could do the same?

Instead of treating every message as isolated, what if every message could access the full conversation history before responding?

So I built conversation memory into SQLite.

Now every message queries the database by phone number and injects the full conversation history into Claude's context before responding.

The result: a bot that actually knows who you're talking to.

The Technical Stack

Why this combination works:

Component	Purpose	Why It Matters
Twilio	WhatsApp API	Reliable business messaging (no Meta API restrictions)
Claude	Conversation AI	Natural, intelligent responses with context awareness
SQLite	Memory storage	Persistent conversation history by phone number
Python+FastAPI	Backend	Webhook handling & lightweight integration

The Architecture

Here's the 4-step flow every message follows:

Receive – Twilio webhook captures the incoming message
Query – SQLite retrieves all previous messages from that phone number
Inject – Full conversation history goes into Claude's context window
Respond – Claude answers based on full context, not isolation

Example SQL query:

SELECT message_text, timestamp FROM conversations 
WHERE phone_number = ? 
ORDER BY timestamp ASC

Result: The bot remembers. The customer feels heard. Sales convert.

The Impact

With conversation memory:
✅ Customers don't repeat themselves
✅ Bot understands context and intent
✅ Sales questions get proper responses
✅ 2-3x higher conversion vs. stateless bots

Without it:
❌ Every question starts from zero
❌ Bot seems broken or rude
❌ Customers leave for competitors

The Lesson

The hardest part of building a WhatsApp bot isn't the code. It's solving the business problem: How do I make customers feel heard?

The tech is secondary. Conversation memory is the bridge between a tool and a real business solution.

Worvi

I built this solution into Worvi — a WhatsApp AI bot you can sell to your clients or use for your own business.

Everything above (Twilio + Claude + SQLite + memory layer) is already built in.

Launching on Product Hunt May 19.

→ Learn more: https://kitbot.gumroad.com/l/worvi-whatsapp

How I built a voice AI agent that answers phone calls and remembers callers

David — Sat, 09 May 2026 18:39:30 +0000

I got tired of seeing small businesses miss calls. So I built Vokio — a voice AI agent that answers real phone calls, remembers callers between sessions, and generates a post-call summary automatically.

Here's the full architecture and how I solved the hard parts.

Stack

Python + Flask (webhook server)
Vapi (telephony + STT)
Claude Haiku (conversation)
Deepgram Nova 3 (Spanish STT)
Azure TTS (Spanish voices)
SQLite (memory between calls)

How it works

Vapi handles the phone call and speech recognition. Instead of using a built-in LLM, I configured Vapi to use a Custom LLM — pointing it to my Flask server. Every time the caller says something, Vapi sends a POST request to my endpoint and expects an OpenAI-compatible streaming response.

@app.route("/chat/completions", methods=["POST"])
def chat():
    data = request.json or {}
    phone = data["call"]["customer"]["number"]
    last_message = [m for m in data["messages"] if m["role"] == "user"][-1]["content"]
    response = generate_response(phone, last_message)
    # return SSE streaming response

The memory system

This is the interesting part. Every caller is stored in SQLite by phone number. When a call comes in, I query the DB and inject the caller's history into the system prompt.

def generate_response(phone: str, user_message: str) -> str:
    caller = get_caller(phone)
    history = get_history(phone)

    if caller and caller.get("name"):
        system += f"\n\nYou already know this caller, their name is {caller['name']}."

    if caller and caller.get("sentiment") == "frustrated":
        system += "\n\nNOTE: This caller was frustrated last time. Be especially patient."

If the caller was frustrated last time, Claude knows and adjusts its tone automatically.

Post-call analysis

When the call ends, Vapi sends a webhook to /end-call. I pass the transcript to Claude and get back:

2-line summary
Urgency score (1-5)
Required action ("call back today", "send quote", "none")
Customer sentiment (satisfied / neutral / frustrated)

All stored in SQLite. The business knows at a glance which calls need attention today.

The tricky parts

Streaming is required. Vapi expects SSE streaming responses. If you return regular JSON the call hangs up immediately. Took me a while to figure that out.

Tool use + verbal response. When Claude uses a tool (like saving the caller's name) without producing text, you get an empty response. I added a follow-up API call to get the verbal response without Claude announcing what it just saved.

dotenv loading. If environment variables already exist in the system, load_dotenv() won't override them. Use override=True.

Result

A 5-minute call costs approximately €0.28. The agent answers in Spanish, detects if the caller switches to English or Catalan, and responds in the same language automatically.

I packaged it as a ready-to-deploy template with 6 business sector prompts included (dental clinic, restaurant, hair salon, real estate, mechanic, hostel).

→ https://kitbot.gumroad.com/l/vokio

Cómo construir un chatbot WhatsApp inteligente con Claude API (Worvi)

David — Wed, 29 Apr 2026 16:44:38 +0000

Cómo construir un chatbot WhatsApp inteligente con Claude API

En este tutorial aprenderás a crear Worvi, un chatbot WhatsApp potenciado con Claude API que puede gestionar conversaciones con memoria y monetizar a través de Gumroad.

1. Requisitos previos

Cuenta de WhatsApp Business API (Meta)
API Key de Anthropic Claude
Python 3.9+
Servidor con HTTPS (Ngrok, Heroku, o VPS)

2. Estructura del proyecto

worvi/
├── bot.py
├── config.py
├── database.py
├── webhook_handler.py
└── requirements.txt

3. Integración con Claude API

Usamos claude-haiku-4-5 para respuestas rápidas y económicas:

import anthropic

def generate_response(user_message, conversation_history):
    client = anthropic.Anthropic()
    messages = [
        {"role": "user", "content": msg}
        for msg in conversation_history
    ] + [{"role": "user", "content": user_message}]

    response = client.messages.create(
        model="claude-haiku-4-5",
        max_tokens=200,
        messages=messages
    )
    return response.content[0].text

4. Webhook de WhatsApp

Configura el endpoint para recibir mensajes:

from flask import Flask, request

app = Flask(__name__)

@app.route('/webhook', methods=['POST'])
def webhook():
    data = request.json
    phone = data['entry'][0]['changes'][0]['value']['messages'][0]['from']
    user_message = data['entry'][0]['changes'][0]['value']['messages'][0]['text']['body']

    response = generate_response(user_message, get_history(phone))
    send_whatsapp_message(phone, response)
    return {"status": "ok"}

5. Base de datos con historial

Guarda conversaciones para contexto:

import sqlite3

def save_message(phone, user_message, bot_message):
    conn = sqlite3.connect('chats.db')
    conn.execute(
        'INSERT INTO messages (phone, user_msg, bot_msg) VALUES (?, ?, ?)',
        (phone, user_message, bot_message)
    )
    conn.commit()

6. Monetización con Gumroad

Ofrece versión premium a través de Gumroad:

🎯 Worvi - Chatbot WhatsApp con IA
€89 - Configuración e instalación personalizada
→ https://worvi.gumroad.com

Incluye:
✓ Setup de WhatsApp Business API
✓ Integración Claude API
✓ Base de datos de conversaciones
✓ 3 meses de soporte

7. Despliegue

Usa Docker + Render o Railway para producción sin mantenimiento.

¿Listo para automatizar tus conversaciones?

Obtén Worvi hoy → €89 en Gumroad