[DMP 2024]: Develop a AI-based conversational automated call center

[MohitNSamagra]

Ticket Contents

Description

Create an AI-based conversational call system with advanced features to facilitate natural language interactions between GPT/an expert database and an end-caller. The system should accept speech input and provide appropriate responses back as speech. Advanced features should enhance the system's ability to comprehend queries, fostering a more nuanced conversation with fluid transitions between AI and human interaction.

Goals & Mid-Point Milestone

Goals

• The system should seamlessly transition between human and AI interaction(seamless rerouting) for a given decision tree
• The system should ask follow-ups to achieve clarity for a given decision tree
• The system should maintain history awareness by automating data retrieval for a given decision tree
• The system should deliver emotionally natural responses
• The system should detect tone and style in voice
• The system should be optimized for time-to-response and should be able to pre-empt long gaps and fill in accordingly for a given decision tree

Setup/Installation

No response

Expected Outcome

No response

Acceptance Criteria

• When asked a query which the system cannot answer, a seamless transition to a human agent should occur, and the system should take back over in case I have another question or if the agent does not respond
• If a vague question is asked, I should be asked follow-ups for clarity until the system gets the information required for it to give me a response
• My conversation history should allow for me to not have to repeat answers to simpler questions asked in the past, reducing the overall time taken
• I should get responses which are emotionally natural with fillers like hmm, right, okay, etc.
• My responses should be tailored according to the tone/style of my voice and the context of my query.
• My waiting time for responses should not be too long for any query as the system should know how to tackle a long wait by inserting relevant content(music, follow up, etc) seamlessly

Implementation Details

• Utilize state-of-the-art natural language processing techniques to implement advanced features
• Integrate decision trees for clarity nudging and automate data retrieval for history awareness.
• Develop algorithms for tone and style detection in voice and enhance empathy in responses.
• Optimize time-to-response by preempting delays and minimizing filler words.
• Implement a tagging system for categorizing responses based on feasibility, determinism, accuracy, relevance to Indian soil, and pricing considerations.
• Conduct thorough testing to validate the functionality and effectiveness of the advanced features.
• Document the development process, including feature implementation details and testing procedures.

Mockups/Wireframes

No response

Product Name

Automated Call Center

Organisation Name

SamagraX

Domain

⁠Service Delivery

Tech Skills Needed

Artificial Intelligence, Natural Language Processing, Python

Mentor(s)

@ChakshuGautam

Category

Data Science, Machine Learning