CallVani

• 

• 

• 

• 

Inspiration

Real-estate teams receive hundreds of leads from websites, property portals, ads, and social media, but many high-intent buyers never receive a fast and consistent follow-up. Sales agents spend hours repeatedly calling leads, asking the same qualification questions, and manually recording responses.

The problem becomes even harder in multilingual markets like India, where customers are more comfortable speaking naturally in different languages.

That inspired me to build CallVani — an AI voice agent that can handle real phone conversations and convert an unqualified inquiry into structured, actionable information for a sales team.

What it does

CallVani is a multilingual AI calling agent designed to automate real-estate lead qualification and follow-up.

When a new property lead enters the system, CallVani can call the lead and naturally collect information such as:

• Preferred city or locality
• Property type
• 2BHK / 3BHK preference
• Approximate budget
• Buying timeline
• Site-visit availability
• Questions or special requirements

Instead of giving the sales team a recording they need to listen to manually, CallVani converts the conversation into a structured lead summary.

High-intent or complex conversations can be escalated to a human agent so AI does not become a barrier when human assistance is actually needed.

CALL-E Integration

For this hackathon, I am extending CallVani with a CALL-E-powered calling workflow.

CALL-E is used to connect the AI workflow with real phone conversations. The goal is to let CallVani initiate a call, conduct the qualification conversation, capture the outcome, and return structured information that can be used by the sales workflow.

This makes CALL-E part of the actual execution layer rather than simply using it as a reference or standalone demo.

How I built it

CallVani is designed as an event-driven AI voice workflow.

A lead triggers the calling workflow, business context and qualification goals are provided to the agent, and CALL-E handles the real-world phone interaction.

After the conversation, the application processes the call result and converts it into structured information that can be displayed in the dashboard or sent to another business system.

The architecture is designed so the same calling workflow can later be reused for industries such as healthcare appointments, customer support, hospitality, recruitment, and local businesses.

Challenges

One of the biggest challenges with voice AI is that real conversations are unpredictable.

People interrupt, change topics, use mixed languages, give incomplete answers, and ask questions that are not part of a predefined script.

Another challenge is making the system collect useful structured information without making the conversation feel like an automated IVR.

The workflow therefore focuses on the business goal of the call rather than forcing every customer through an identical script.

Accomplishments

I am particularly proud of turning CallVani from a general AI calling concept into a focused real-world workflow where the result of the phone call can directly help a sales team take its next action.

The project demonstrates how autonomous phone agents can do more than simply answer questions — they can perform useful business work and return actionable results.

What I learned

Building CallVani reinforced that successful voice agents need much more than speech generation.

They need clear goals, business context, conversational flexibility, reliable structured outputs, and a safe path to human intervention.

CALL-E provides an interesting layer for connecting AI agents with real-world phone interactions.

What's next

Next, I want to expand CallVani with:

• More Indian languages
• CRM integrations
• Automatic lead scoring
• WhatsApp follow-ups after calls
• Appointment and site-visit scheduling
• Human-agent handoff
• Inbound and outbound workflows
• Analytics for call outcomes and conversions

The long-term vision is to make CallVani a multilingual AI voice workforce for businesses that still depend heavily on phone calls.

Built With

• agents
• ai
• api
• call-e
• conversational
• llm
• mongodb
• node.js
• react
• rest
• typescript
• voice
• webhooks