SmartRent - Empowering users to take control of their car rental experience.

Design a conversational AI assistant for a car rental platform that could intelligently guide users through different rental scenarios - from quick business bookings to complex family vacation planning - while maintaining natural conversation flow and reducing booking abandonment.

Discovery → User research, journey mapping, persona definition
Design → Conversation flow architecture, prototype
Validate → User testing, stakeholder feedback

Services:

Sector:

Car Rental

Team:

Copilot, Claude (protopersonas), Miro (journey mapping), Figma (conversation flow prototypes)

Challenge:

Tools used:

4 months

Duration:

1 Product Designer (me), 2 Developers, 1 Product Lead, 1 Information Architecure

Proto-Persona Development Process

Given project timeline constraints, we adopted a strategic proto-persona approach leveraging existing knowledge and AI assistance:

Data Sources:

• Project Lead's domain expertise and user insights

• Historical booking pattern analysis

• Competitive analysis of rental behaviors

AI-Assisted Research Method:

Used Claude and Copilot to synthesize available data into structured proto-personas, helping us:

• Identify behavioral patterns from existing user data

• Generate hypotheses about user motivations and pain points

• Create detailed persona narratives based on real usage patterns

• Validate assumptions against industry best practices

Discovery & Strategic Research

Profile Rental Scenario Example

🥸 Conversational profiles:

• Educational Support (New Driver) - Requires explanation and reassurance

• Comparative Guidance (Family Travel) - Needs detailed comparisons and recommendations

• Simplified Navigation (Senior) - Benefits from patient, clear communication

• Efficiency Focus (Business Traveller) - Values speed and business-oriented features

Insights & Behavioral Patterns

👉 Key Behavioral Patterns Identified:

• Age-based communication preferences: Younger users accept a casual tone, seniors prefer a formal approach.

• Context-driven information depth: Family trips require more detail than business bookings.

• Insurance sensitivity varies significantly: Young drivers need explanation, business travelers need efficiency

• Technology comfort levels: Younger users handle complex interfaces easily, while seniors need simpler, step-by-step guidance and quicker access to human support

Sample of profiles, flows, and fallback scenarios

👉 Marina, 35, planning a family vacation to Orlando with partner and two children (ages 3 and 7)

👉 Trip: 7-day vacation, arriving at Orlando airport

👉 Concerns: Safety for kids, luggage space, budget management, and entertainment for long drives

Scenario Context

Learnings and next steps

💡 What I Learned

• Designing for multiple user profiles doesn’t always mean multiple bots — sometimes, the best experience comes from simplicity and consistency, especially when dealing with generative AI and automation.

• Collaboration with stakeholders is crucial when business logic challenges user expectations — aligning the design with both needs is where real impact happens.

• The visual design, including subtle typographic details, has a significant impact on user navigation and understanding in conversational interfaces.

🌀Design Trade-offs & Strategic Decisions

• While the initial proposal aimed to adapt tone and flow based on each user profile, we chose to consolidate the conversational tone to a more neutral and direct style, ensuring the same bot experience could serve diverse user needs without confusion.

• Business constraints required us to prioritize a flow start before recommending vehicle options. This sequencing reduced early personalization but ensured the solution aligned with operational requirements and system integrations.

• Emphasizing key actions in bold text was a lightweight but impactful tactic to improve clarity, especially for first-time or less tech-savvy users.

📈 Future Metrics & Optimization Plan

Although I transitioned off the project before we conducted quantitative validation. We had outlined some next steps to measure and optimize the experience:

• Conversation Completion Rate: Percentage of users who complete the booking process through the assistant without escalation to a human agent.

• User Satisfaction Score: Post-interaction feedback using a quick thumbs-up/thumbs-down or a 5-point scale.

• Fallback Effectiveness: Measure how often fallback flows recover the conversation versus ending in failure or hand-off.