Genie experience center - Designing a voice-based agent to inform and engage leads during their waiting time at real estate site visits.
Voice agent
.
Web app
.
Sep 2024 - Nov 2024
My role
End to End Product & Interaction design
Team
1 PM, Business heads, Operations team & 2 Engineers
Time line
Sep 2024 - Nov 2024
Background context & Problem
In Anarock's mandate model real estate projects, weekend site visits experience a significant surge in footfall, leading to long waiting times of 45 minutes to 1 hour for prospective buyers. This waiting period results in disengaged leads and missed opportunities
Solution
🎙
Interactive voice-based AI to engage leads during long weekend wait times
🌍
Multilingual & conversational AI sharing project info — pricing, amenities, and offers
😄
Fun & engaging — turns idle time into a branded interaction
📊
Captures buyer intent & improves site visit efficiency
Observing & Understanding environments
Since this was a voice-based interaction, I focused on key questions around:
👩🏻💻
Environment at the site & interaction between agents & leads
Understanding the real-world conditions—noise levels, user behavior, and flow of people during site visits.
🔄
Placement of the devices & how many devices will be required
Where should the devices be placed for maximum visibility and usability? How many will be needed to handle peak crowd flow?
🚀
Tech feasibility for noise & far speech recognition
Exploring the technical possibilities for accurate voice recognition in noisy, far-field, multi-user environments.
On ground learnings
To understand these factors better, I visited the site as a lead and observed the environment
key insights & Challenges need to be solved for
To understand these factors better, I visited the site as a lead and observed the environment
🎯
Conversations must stay focused on project
While users may ask general questions, the agent should smartly guide the conversation back to project-related info
📊
Capturing valuable buyer data
The system must subtly collect key user inputs (like budget, interest, preferences) without feeling intrusive.
📊
Capturing valuable buyer data
The system must subtly collect key user inputs (like budget, interest, preferences) without feeling intrusive.
🚀
Tech feasibility for noise & far speech recognition
Exploring the technical possibilities for accurate voice recognition in noisy, far-field, multi-user environments.
Final designs after multiple iterations
Scroll to see complete flow >>>
Home screen where all the live ongoing matches and popular players stocks are displayed
Users can engage in fun banter with AI-generated stickers acc to match situation
Cross selling stocks of players during peak moments
Stocks buying flow
More visulas
Post-Launch Success Metrics
Post-launch, success will be measured by how actively users engage with the banter feature during live matches, the frequency of interactions during peak moments, and the uplift in player stock transactions triggered by in-banter prompts. These insights will help us enhance fan retention, deepen emotional engagement, and drive more contextual monetization.