Universal designer of smart AI assistants with call script settings
August 2, 2024
4 minutes read
SUMMARY
Our client for this project is a call center company. They have many repetitive scenarios that they are approached with, our task was to solve the issue of automation
We created a Web application completely from scratch, which allows you to connect your phone's Trunk, create and edit call scripts. Where you can set up integrations with third-party APIs, save responses as variables, create dynamic robot responses in real time
Integration with WhatsApp and SMS has been implemented, they can be configured in the general settings of the call script, duplicate the AI assistant's responses with their help, and much more
TECH STACK
Asterisk
Digital Ocean
GPT
Docker
Next-js
Django
OpenAI-Text-To-Speech
Open AI Speech-To-Text
DELIVERY TIMELINE
1 week
Solution Architecture Design
2 week
Development of Real-time module TTS and STT
1 week
Basic UI/UX development
2 week
Development of Asterisk module and main backend
2 week
Creating a client part using Next.js
1 week
Deployment & Testing
TECH CHALLENGE
Automation of Repetitive Call Scenarios: One of the primary challenges was developing a robust system to automate the repetitive scenarios faced by the call center, ensuring that the Web application could handle various use cases effectively.
Custom Trunk Integration: Integrating the client's phone Trunk system into a custom-built Web application required careful handling of telecom protocols and ensuring seamless connectivity for uninterrupted call management.
Real-time AI-driven Responses: Implementing real-time AI-driven responses within call scripts posed a significant challenge, as the system needed to generate accurate and contextually relevant responses on the fly.
WhatsApp and SMS Integration: Seamlessly integrating WhatsApp and SMS into the call scripts was essential for duplicating AI assistant responses, requiring complex API integrations and real-time synchronization.
Dynamic Call Script Management: Designing a flexible interface for creating and editing call scripts in real-time, with the ability to integrate third-party APIs and save responses as variables, was crucial for enabling dynamic interactions during calls.
SOLUTION
We leveraged Django to build a backend that manages and automates repetitive call scenarios. By utilizing Asterisk, we integrated it with Django to handle telephony operations, allowing the Web application to execute predefined call scripts automatically. This setup ensured seamless handling of repetitive tasks with minimal human intervention.
Integration of the client's phone Trunk was achieved using Asterisk within a Docker container for an isolated and secure environment. This allowed us to manage the telephony functions directly from the Django application, ensuring smooth and reliable Trunk connections for all call operations.
For creating and editing call scripts, we utilized Next.js to build a responsive and intuitive front-end interface. Django served as the backend, handling the logic for dynamic script creation, API integrations, and variable management. We designed the UI/UX in Figma to ensure a user-friendly experience, enabling real-time updates and edits to call scripts with seamless backend communication.
By integrating the OpenAI API into the Django backend, we enabled real-time AI-driven responses within the call scripts. The system could process input data, generate contextually relevant responses, and send them back through the call interface with minimal latency, all within the Asterisk telephony environment.
Using Django, we built a flexible integration layer that connects to third-party APIs for WhatsApp and SMS. These integrations were configured within the call script settings, allowing the system to duplicate AI assistant responses via WhatsApp and SMS in real time. Docker was employed to manage these services in a containerized environment, ensuring consistent and scalable performance.
The entire application stack, including Django, Asterisk, and the AI integrations, was deployed using Docker containers, ensuring scalability and easy management of the system. This architecture allowed us to quickly scale the application to handle increased loads while maintaining high reliability, even during peak call volumes.
.svg)
