Revolutionizing Voice Call Management with AI during DND Mode: An Innovative Solution to Modern Communication Challenges
Managing incoming calls efficiently is vital for maintaining productivity and ensuring effective communication in the digital age. However, traditional call management systems often need to catch up in two critical areas: conveying unavailability to callers during Do Not Disturb (DND) mode and understanding the purpose of their calls.
This case study explores how I identified the pain points, ideated a solution, designed and prototyped an AI-driven system specifically for DND mode, and planned the next steps for integrating this solution into mobile operating systems.
Don’t miss the live prototype(Developed version) at the end 👇.
Pain Points in Current Call Management Systems
Current call management systems often fall short in two critical areas, particularly during DND mode:
— Communicating the receiver's unavailability to callers while in Do Not Disturb(DND) mode.
Unlike visual indicators used in platforms like Microsoft Teams and Google Chat, which show “Out of Office” or “Busy” statuses
Phone calls lack a similar non-visual mechanism to inform callers of the user’s unavailability, making it challenging for callers to understand the user’s current status.
leading to repeated call attempts and interruptions.
— Understanding the purpose of their calls on the de-active of DnD.
If you receive n calls without answering them you will not get the purpose of the call.
As there are different modes in DND like
- Allow calls only saved contacts
- Allow calls only from favourite contacts
- Allow all calls.
The receiver might not know whether it’s an important call to attend.
User research
I conducted a small user research to determine how the users understand the call's purpose and prioritize reverting to the calls when they set their phone to DND.
“Consider you’re in an important meeting with your phone on DND mode and you have received 10 calls. Some from saved contacts and some from unknown contacts/new numbers.
Which contact would you revert first and how would you decide which call to revert first?
”
Findings
- 99% — mostly revert back to saved contact calls first.
- For not saved numbers they will check with Truecaller for spam.
- If they receive repeated calls from an unknown number they verify in Truecaller that if it is not spam they classify it as an important one and revert back.
Users often miss important calls because they need help discerning the urgency or purpose of the call.
Ideation: Addressing the Pain Points
Given these pain points, I embarked on an ideation journey to brainstorm potential solutions. The goal was to create a system that:
- Customized unavailability status to the caller.
- Ensures important calls are noticed with categorisation of the calls based on the conversation.
- Allows for the prioritization of calls based on their urgency.
- Provides multilingual support for better communication
- Effectively manages spam and unwanted calls
Market Research
To better understand the landscape of AI-driven call management solutions, I looked into existing technologies and their limitations:
- Feature: Google Duplex, demoed during Google I/O, allows Google Assistant to make calls on behalf of the user to book appointments or make reservations.
- Limitation: Google Duplex is designed for specific tasks like booking appointments, call screening and requires the user to manually initiate the call. It do operate autonomously but not all the calls, meaning it can only automatically screen calls unknown calls, first time callers. This process limits its effectiveness in scenarios DND mode, screening saved contacts and also provide a custom message set by users in real time.
- Duplex also doesn’t handle situation where user is not available to take the call.
Samsung Text Call
- Feature: Similar to Google Duplex, Samsung’s Text Call feature allows users to screen calls by converting the caller’s voice into text and displaying it on the screen.
- Limitation: Like Google Duplex, it requires the user to actively engage with the call screening process. This does not serve the purpose during DND mode, as the receiver might not receive the call or cannot initiate the call screening feature to know the call details.
Both of these technologies offer impressive capabilities but fall short of providing a seamless solution for users in DND mode with both saved and unknown contacts.
This gap highlighted the need for an autonomous system that can manage calls without requiring real-time user interaction handling both saved and un saved contacts.
Design and Prototyping
Design Phase
In designing the solution, prioritizing the user experience and seamless integration with existing mobile systems.
This AI Assistant feature has to be available only if the phone is set to DND mode. So when a user tries to enable it without DND mode, the app should prompt for enabling DND.
Customizable Status Messages
Users can set personalized messages that the AI conveys to callers, explaining their unavailability during DND mode and only when the automated call answering feature is turned on.
Conversation History
Users can review the AI-caller interaction, including a summary of the conversation, to make informed decisions about which calls to return.
Solution to Address Non-Visual Communication
To address this, my AI system leverages voice-based notifications to inform callers of the user’s status. For example:
- Status Announcement: When a call is received during DND mode, the AI answers the receiver's status based on the message configured by the receiver
“Hi, I’m AI Agent answering for the receiver. He is currently busy right now with meetings.” - Purpose Inquiry: The AI then asks the caller to state the reason for their call
“Please let me know why you’re calling, and I’ll make sure to notify [User’s Name] when they are available.” - Categorization: Based on the caller’s response, the AI categorizes the call and logs the conversation details for the user to review later.
Live Prototyping(Developed version)
To truly test the solution, an implemented/developed prototype was essential. Unlike design prototypes, which rely on pre-defined responses and cannot simulate the dynamic nature of actual phone conversations, a developed prototype allows me to handle the numerous edge cases that arise in real-world usage
I proceeded to develop the app and AI agent. It’s demo go as below.
It’s not the end. Edge cases like no input, call ended abruptly, abusing or harmful situation are also handled during the development.
Currently, this developed prototype is limited to non-IVRS calls and limited to the English language.
Next Steps
With the successful implementation of the live prototype, the next steps involve proposing the integration of this AI-driven call management system into mobile operating systems. This will ensure that users can benefit from seamless call management during DND mode, enhancing productivity and communication efficiency.
Proposed Integration Steps
User Feedback and Iteration: Conduct extensive user testing and gather feedback to refine the AI interactions and improve the overall user experience.
Scalability and Multilingual Support: Ensure the system can handle a large number of users and provide support for multiple languages to cater to a global audience.
Integration with OS(mobile): Currently this is a simulation, but going forward this can be integrated into the Mobile operating system to bring out a more seamless on-device experience.
Android is building on the Gemini Nano(on-device) AI model which was currently in private preview on fewer devices (pixel 8, Samsung S24, S23).
During the recent WWDC 2024 Apple.Inc. announced the ChatGPT integration into SIRI(voice assistant for Apple Ecosystem).
So with extensive user testing and improving the features if this can be taken into the OS that will be much more helpful to the larger set of users.
Hurray you have made it to the end of the case study. Hope you liked it.
This was just a trailer of the project I have done in 2 weeks of span.
I appreciate your feedback on this article. Your thoughts and suggestions help me improve my content. Thank you! 😊👍
Interested in how I developed the live prototype? let me know in the comments so that I can make a tutorial of it.
Happy Designing✏️ & Developing 🧑💻.