AsSIXTant

Inspiration

We all have probably experienced this before when going on a trip with a rental car: Arriving at the airport, long line at the counter (especially at Sixt because everybody wants a Sixt) and the vehicle you chose is probably not available... Wouldn't it be nice to have a smooth digital solution to assess your options for you? Thats what we thought as well! So, not to worry, the Assixtant will assist and guide you in the process, maybe you want to opt for a nice Ford Mustang convertible instead of that basic Volkswagen for the trip with your friends to spice things up a little bit?

What it does

Assixtant is an AI-backed Assistant with a nice looking 3D Avatar that assists you in the process of finding a suitable car and the best coverage and add-on options for your trip. It takes your personal preferences into account, can edit your booking during the conversation and makes the whole process of choosing the right vehicle for your purpose smooth and easy.

How we built it

We used Python for our backend, connected our service to the Google Gemini API and made use of the very powerful Microsoft Azure Text-to-speech and Speech-to-text models, which also provide the visemes for our 3D avatar. For the frontend we used Flask and a 3D Avatar, that was created in Blender. We split the work across our three team members, each attending to one of the topics: backend and assistance logic, TTS/STT implementation and frontend avatar

Challenges we ran into

Getting the Avatar to work properly with visemes and our speech-to-text approach was challenging. However, we are quite happy that we accomplished to make it work in such a short time and are quite pleased with the results.

Accomplishments that we're proud of

Building a working solution like this in effectively only one day, we are quite proud to have managed that with a team of three :)

What we learned

We learned a lot about the powerful capabilities that services like Google Cloud and Microsoft Azure provide. That enables us to build cool tools like that from scratch in a short amount of time. Also, we built a new team in this year's hackatum, so we also got to know each other and are happy to have founded the new Assixtant team.