🧠 About the Project — Blip Voice 🌟 Inspiration

Businesses in India heavily depend on phone calls for sales, customer support, and follow-ups. However, many small and medium businesses still struggle with:

Missed calls

Untrained staff

Poor customer communication

No proper follow-up system

No call recordings or analytics

This repeated loss of leads and customers inspired us to build Blip Voice — an AI system that makes real phone calls on behalf of businesses and ensures no customer is ever lost due to poor communication.

🏗️ How We Built Blip Voice

We built Blip Voice using a combination of AI, low-code automation, and voice technology:

Frontend (Web App)

Built using Base44 / Lovable

Pages created:

Agent Builder

Contact Upload

Campaign Launcher

Call Results Dashboard

User Profile

Authentication: Google + Email/Password

Backend (Automation Engine)

We designed modular workflows using n8n, including:

Assistant creation workflow

Contact management workflow

Campaign calling workflow

VAPI event listener workflow

Call results storage workflow

AI Voice Calling

We used VAPI AI to:

Create intelligent voice assistants

Place outbound calls

Receive events like:

call.started

transcript.updated

recording.available

call.completed

Data Storage

Used n8n Data Store as a mini-database

Stored:

callId

transcript

summary

recording URL

call outcomes

lead status

Summaries & Analysis

Call summaries were generated using OpenAI models. We formulated a simple scoring mechanism:

Engagement Score

𝛼 ⋅ Call Duration + 𝛽 ⋅ Keywords Matched Engagement Score=α⋅Call Duration+β⋅Keywords Matched 🎓 What We Learned

Building Blip Voice taught our team several valuable lessons:

How to integrate real-time AI calling systems

How to use event-driven architectures

How to capture and process call transcripts

How to structure prompts for multi-role AI agents

How to store and manage data using n8n Data Stores

How to build a full working MVP using low-code tools

How to design scalable workflows and pipelines

How to collaborate effectively as a 4-member team

🧩 Challenges We Faced

VAPI Webhook Configuration

We faced issues where VAPI didn’t send call events due to:

Incorrect webhook URL

Assistant not saving serverUrl

PATCH API returning 404

We solved this by debugging API base URLs and verifying webhook activity logs.

Handling Real-Time Call Events

Managing events like transcript.updated required careful workflow design to avoid:

Overwriting transcripts

Partial event loss

Duplicate entries

Low-Code Integration Complexity

Although low-code tools reduce development time, combining:

Base44

n8n

VAPI

required multiple iterations to ensure everything synced properly.

Time Constraints

Building a full AI-powered calling platform in a short hackathon time frame required dividing tasks efficiently among team members:

Frontend

Backend workflows

AI calling logic

Testing & debugging

🚀 Conclusion

Blip Voice is our attempt to create an affordable, scalable, and intelligent calling system for businesses in India. This project showed us the power of combining:

Voice AI

Automation

LLMs

Low-code tools

We are excited to continue improving Blip Voice and explore its potential for the future of business communication.