Hey Alexa — voice based chat

a conversational user interface concept for Airbnb Hosts and Guests
This project was a research work done at CMU to reimagine the future user experiences at Airbnb as we navigate into the next era of computing.
I designed a voice based conversational agent (CUI) for Airbnb.
COMPANY
Airbnb
(Concept)
Timeline
2 months
WHAT I DID
Research
Script Design
Prototyping
Physical Prototype
Tools
Figma
VoiceFlow
PLATFORM
iOS App
Web
INTRO

Imagine if you could just record a video of your place and Airbnb automatically writes your listing, pick photos, set prices, and even give you tips based on guest feedback.

Hosts need to list their properties on the Airbnb Platform using a long manual process. They are in rush & do not possess the marketing skills to do it in the best way possible. Generative AI makes it easy to do marketing & do listing in the best way possible.
USER PROBLEM

The Guests have many disagreements/questions which they are not able to ask due to limited time and slow response rate.

"The listed rate for the property is $131/night, yet the Host asserts it should be $171/night. Upon contacting Airbnb Customer Service, they declared that it's Hosts responsibility to correctly list the amount" - Airbnb Guest
SOLUTION

A Conversational interface for Hosts

I mapped out ways to address these questions and guide users to specific answers that don’t compete with other essential information on our platform.
I started crafting diagrams, charts, sketches of the conversation flow
Fig 1. Proposed solution
context

Absence of Airbnb Hosts leads to confusion and dissatisfaction among Guests

Hosting on Airbnb can be demanding, especially for hosts who manage multiple listings or have other personal or professional commitments. It requires time and effort to handle bookings, inquiries, guest interactions, cleaning, and overall management of the hosting process.
Major identified issues:
  1. Property Listing:
Hosts suffer from correctly listing & updating their preferences.
  1. Hosting Experience:
Some Guests need constant attention, especially in urgent situations.
  1. Feedback:
Hosts need regular feedback to know Guest's preferences.
approach

The project was open ended and I defined a direction that I would take to find user problems to solve using this technology

Understand the technology & its limitations.
Find key stakeholder's & their problems.
Map the User Journeys & business opportunities.
design thinking

What problems can Conversational AI solve ?

Rather than the traditional design process, I followed upside down process, where I investigated how Conversational AI technology can add value to the lives of Airbnb customers.
identifying users

To identify the key stakeholders, I decided to create a stakeholder map and their relationships

Fig 2. Stakeholder Map helped us to identify 'Hosts' as the customer and its impact. My research dealt into looking into possibilities of engaging AI experiences for multiple stakeholders, To narrow down the scope, I decided to focus on the Guest & the Host experience.
journey mapping

I mapped the AI opportunities at every touch point for Guests & Hosts

Fig 3. To identify the seamless user experience, I created a journey map & mapped the opportunity at each touch point. Later, I decided to focus on the "Arriving & Staying" zone.

I tested multiple happy path edge cases with ChatGPT to map the conversations on VoiceFlow

Fig 4. Working session sketches and notes for the happy path scripting. The process to check the edge cases and breakpoints using ChatGPT was helpful before the user testing sessions and saved us alot of time.
other opportunities

A Conversational interface for Guests

When texting someone, how they look or sound is not as apparent as what they’re thinking. In that sense, it’s not hard to imagine a model where we treat bots and humans alike (at least in chats and messaging apps). 
Fig 5. A virtual agent that utilizes generative ai to enhance living experience for Guests. Prototyped with VoiceFlow
CONCLUSION

Learnings & Failures

It is the intuitiveness and clarity of the solution that drives the product. Some projects start with clarity of the solution instead of problem.
It is hard to discuss futuristic ideas with target audience if the fidelity of the solution is not correct. I learnt through the project that it was initially difficult for the Hosts to understand our idea, but after watching the video, they could imagine themselves into the situations where this can be useful to them.
How to balance engineering effort vs innovation? In order to test the viability of the idea, I did a social test and posted it on Linkedin to see how Linkedin audience would react to it, and if it is hard to imagine such a solution.
— TOP
— Core Experience
— Bottom
< LAST
Ask Dave — Honda
NEXT >
Microsoft Teams

Last project

Contact me

Next Project