Designing AI-powered messaging platform for business
Transitioning Milestone’s customers to a Hybrid Human-AI Support Model
Transitioning Milestone’s customers to a Hybrid Human-AI Support Model
The Solution
The All-in-One Messenger App
I built 2 features, in addition to a basic messenger:
Route messages to the right team
This feature ensures that incoming messages are directed to the appropriate team or team member within an organization. I designed a global setting for this.
Engage customers at every stage
This feature focuses on maintaining customer engagement throughout their entire journey with the business. Collect payments. Build rapport with appointment reminders. Send review, referral, and survey requests.
The Challenge
Simplifying chatting with customer support
As chat-based customer support becomes the norm, I worked with Milestone to simplify how businesses handle multiple messaging channels. I developed a customer support product that facilitates instant responses across multiple channels.
As chat-based customer support becomes the norm, I worked with Milestone to simplify how businesses handle multiple messaging channels. I developed a customer support product that facilitates instant responses across multiple channels.
4 most common entry points where customers need support
Team
1 Product Design Lead (me)
2 Product Managers
6 Engineers
2 QA
2 Product Marketing
1 Product Design Lead (me)
2 Product Managers
6 Engineers
2 QA
2 Product Marketing
1 Product Design Lead (me)
2 Product Managers
6 Engineers
2 QA
2 Product Marketing
Timeline
2 Months
2 Months
2 Months
Responsibility
Research
Strategy
Design
Prototyping
Project Management
Research
Strategy
Design
Prototyping
Manage Project
Research
Strategy
Design
Prototyping
Project Management
Results
65% adoption of this product by Milestone’s customers
Response Time: 1 minute average response time for businesses to customer inquiries.
The Solution
The All-in-One Messenger App
I built 2 features, in addition to a basic messenger:
Route messages to the right team
This feature ensures that incoming messages are directed to the appropriate team or team member within an organization. I designed a global setting for this.
Design
A New feature emerged: AI Summarization
After concept testing with 10 participants, a new need surfaced: the desire for a summarization feature within our platform. Messages tended to be long and lose context. Agents expressed the necessity for a tool that could condense conversations efficiently.
Summarize conversations in seconds
Access AI-generated summaries to quickly recap conversations across channels, share them with colleagues to collaborate and solve problems in real-time.
Concept 1:
The AI summarizes the screen with all cards collapsed. Users can select a card to expand and view its details.
Concept 2:
The AI summarizes everything in one area and the user can ask questions to get more details.
Although Concept 1 was the clear winner, AI Summarization still needed to be improved
I conducted usability testing with 8 users.Concept 1 was the clear winner because it made it easy for the user to see all the highlighted points. However, users struggled to differentiate recent vs. older summaries, so I introduced "Recent" and "Older" groups with dates for clarity. Tags were also added to specify section requirements.
User Interviews
Engage customers at every stage
This feature focuses on maintaining customer engagement throughout their entire journey with the business. Collect payments. Build rapport with appointment reminders. Send review, referral, and survey requests.
Discovery
Shift from AI-driven to hybrid
human-AI support
Initial AI Reliance:
Initially, I worked on an AI-driven approach to responding to customer support messages. User interviews revealed a strong preference for human interaction over purely AI responses.
Pivot:
I helped transition the product direction to a hybrid model where AI handles routine tasks and human agents manage complex interactions, blending efficiency with a personalized touch.
Discovery
Initial Prioritization and Deprioritization
To facilitate this transition, I built a product roadmap:
Deprioritized Aspects
Advanced AI Features: Postponed predictive text and sentiment analysis to ensure timely market entry.
Extensive Customization: Deferred deep UI and workflow customization to focus on core functionalities.
Prioritized Features
Basic Messenger Capabilities: Prioritized integration with WhatsApp, Instagram, Facebook, Google Business, Apple Messaging, and SMS.
Triaging and Message Routing: Developed efficient systems for agent assignment, conversation management, and automated responses.
To facilitate this transition, I built a product roadmap:
Deprioritized Aspects
Advanced AI Features: Postponed predictive text and sentiment analysis to ensure timely market entry.
Extensive Customization: Deferred deep UI and workflow customization to focus on core functionalities.
Prioritized Features
Basic Messenger Capabilities: Prioritized integration with WhatsApp, Instagram, Facebook, Google Business, Apple Messaging, and SMS.
Triaging and Message Routing: Developed efficient systems for agent assignment, conversation management, and automated responses.
Discovery
A New feature emerged: AI Summarization
After concept testing with 10 participants, a new need surfaced: the desire for a summarization feature within our platform. Messages tended to be long and lose context. Agents expressed the necessity for a tool that could condense conversations efficiently.
Summarize conversations in seconds
Access AI-generated summaries to quickly recap conversations across channels, share them with colleagues to collaborate and solve problems in real-time.
Concept 1:
The AI summarizes the screen with all cards collapsed. Users can select a card to expand and view its details.
Concept 2:
Concept 2:
The AI summarizes everything in one area and the user can ask questions to get more details.
Concept Testing
Although Concept 1 was the clear winner, AI Summarization still needed to be improved
I conducted usability testing with 8 users.Concept 1 was the clear winner because it made it easy for the user to see all the highlighted points. However, users struggled to differentiate recent vs. older summaries, so I introduced "Recent" and "Older" groups with dates for clarity. Tags were also added to specify section requirements.
After Usability Testing
The final concept
After two months of diligent work, We successfully launched its All-in-One Messenger App with AI Summaries.
I built out the final concept, and shipped it! I focused on streamlining the workflow with AI handling the heavy lifting.
Outcome
Messenger hits 65% adoption within 6 months
65% adoption of this product by Milestone’s customers
1 minute average response time for businesses to customer inquiries.
85% of customers were now using the "auto-assigned" feature, instead of the "basic" assignment.
65%
Adoption
1 min
Response time
85%
Using new features
Retrospective
Future Considerations:
Future Considerations
To ensure continued improvement, I left the team with 2 future concepts to consider:
Write better replies with AI
AI-writing assistant to rephrase your responses with more professional language and better grammar for more effective customer communication
Predictive Customer Support:
When a customer has been browsing through support articles about a certain feature, the AI could step in proactively, offering helpful tips or troubleshooting steps. This not only prevents potential issues but also boosts the overall customer experience.
Design
The Final Concept
I built out the final concept, and shipped it! I focused on streamlining the workflow with AI handling the heavy lifting.
Outcome
Messenger hits 65% adoption within 6 months
65% adoption of this product by Milestone’s customers
1 minute average response time for businesses to customer inquiries.
85% of customers were now using the "auto-assigned" feature, instead of the "basic" assignment.
65%
Adoption
1 min
Response time
85%
Using new features
65%
Adoption
1 min
Response time
85%
Using new features
Retrospective
Future Considerations
To ensure continued improvement, I left the team with 2 future concepts to consider:
Write better replies with AI
AI-writing assistant to rephrase your responses with more professional language and better grammar for more effective customer communication
Predictive Customer Support:
When a customer has been browsing through support articles about a certain feature, the AI could step in proactively, offering helpful tips or troubleshooting steps. This not only prevents potential issues but also boosts the overall customer experience.