AI Voice Bot Development for Inbound / Outbound Calls
Budget: €250 – €750 EUR
Project Overview:
We are developing an AI-based calling script using Python, focusing on creating a voice bot for outbound B2B calls.
Current Setup:
Files in Local Directory:
A Python script
A .env file for environment variables
Text files named:
"response" for logging call outcomes
"numbers" presumably for storing phone numbers to call
"data" containing information about companies or products
"botprompt" for predefined bot interactions or scripts
System Architecture:
Speech-to-Text (STT): We will use the Deepgram API to convert speech to text.
Language Model (LLM): For processing and generating responses, we'll utilize the OpenAI API with the gpt-4o-mini model.
Text-to-Speech (TTS): We will convert text back to speech using the OpenAI API with the "Sage" voice.
Call Process:
Outbound Calls:
Calls will be initiated using the Twilio API.
When a call is made, the bot will engage with the recipient, using data from the "data" file to tailor conversations about specific companies or products.
Post-Call Actions:
Logging: After each call, we will record:
The phone number that was called
A brief summary of the conversation
The status of the call (e.g., Lead, Potential Lead, Customer, Not Interested, Did Not Pick Up) in the "response" text file.
Focus:
Initially, we'll concentrate on outbound calls, with the system designed to potentially handle inbound calls in future iterations.
This setup aims to automate and personalize B2B outbound calling, leveraging AI for more effective communication.
We are developing an AI-based calling script using Python, focusing on creating a voice bot for outbound B2B calls.
Current Setup:
Files in Local Directory:
A Python script
A .env file for environment variables
Text files named:
"response" for logging call outcomes
"numbers" presumably for storing phone numbers to call
"data" containing information about companies or products
"botprompt" for predefined bot interactions or scripts
System Architecture:
Speech-to-Text (STT): We will use the Deepgram API to convert speech to text.
Language Model (LLM): For processing and generating responses, we'll utilize the OpenAI API with the gpt-4o-mini model.
Text-to-Speech (TTS): We will convert text back to speech using the OpenAI API with the "Sage" voice.
Call Process:
Outbound Calls:
Calls will be initiated using the Twilio API.
When a call is made, the bot will engage with the recipient, using data from the "data" file to tailor conversations about specific companies or products.
Post-Call Actions:
Logging: After each call, we will record:
The phone number that was called
A brief summary of the conversation
The status of the call (e.g., Lead, Potential Lead, Customer, Not Interested, Did Not Pick Up) in the "response" text file.
Focus:
Initially, we'll concentrate on outbound calls, with the system designed to potentially handle inbound calls in future iterations.
This setup aims to automate and personalize B2B outbound calling, leveraging AI for more effective communication.