Inspiration
We wanted to create a website to make it easy to navigate safely. So we created ‘AMI’ (French for ‘friend’) to help women and non-binary people walk around while helping to keep them safe. Walking alone, especially at night, can be a daunting experience and for many women and non-binary people requiring people to think about many factors. We wanted to simplify this process and have everything you need in one place, including safe routes to take, an agent who you can call anytime, and a supportive network of other women and non-binary people.
What it does
Our project allows people to see nearby events and register. The safest routes to navigate around using streetlights, open shops, and other users ratings. Each user can rate a route they have taken and identify any safety aspects they want to highlight. We also set up an AI agent who you can call to make it seem like you’re not alone. On user profiles you can add your interests and emergency contacts. Our buddy system allows verified tier 2 users to match with other users of similar interests so they can chat and also go to events together. To access tier 2, the user will have to go to one event where they can get verified as a trusted user. This is just to block out anyone pretending to be someone else from direct messages and meeting up with other users outside organised events. Additionally tier 2 users ratings of routes are valued higher as we know they are more trustworthy.
How we built it
The main AI tools we used were GitHub co pilot and gemini through antigravity, we came together as a group and drafted up a detailed plan so we can make specific and well defined prompts to our ai agents.
We created a GitHub repository to ensure everyone in the team could contribute their local work and access any new updates easily no matter time and location.
The main languages of the product are; Typescript for the backend and apis, and html and css for the frontend/ui. PostgreSQL was used to design the database which will manage user information, event locations, ai call features etc.
Challenges we ran into
We ran into trouble accessing a working map because we used the google map open API first and tried to draw paths on top of it for our website We overcame this by switching to open streetmap.
We had a lot of issues with running out of credits for free trials so we continuously switched which laptop was doing each thing to max out our usage.
Integrating our ElevenLabs AI agent successfully.
Accomplishments that we're proud of
Our main struggle was working on how to seamlessly integrate our eleven labs agent “Adam” - when the call button would work, the voice response would stop and vice verse. But the moment everything worked together in tandem with the ai voice we were especially proud.
We were also proud of managing to complete so much in so little time regardless of half the team not having computer science/tech background, fortunately with the help of ai and our creative minds we managed to build a full working web application from scratch.
What we learned
Some of us had previously had some prejudice against ai due to their effects on the planet and young minds but whilst working on this project and also during the wise panel talk with their employees we managed to see that ai is beneficial in multiple ways when used wisely.
We also learnt how to work well in a team with members from different backgrounds, degrees and universities whilst successfully communicating with each other and asking for help when needed.
What's next for AMI
Our next goal is to integrate the Presage camera tracking into the video call because we had challenges incorporating with the ElevenLabs model. This would measure the user’s breathing rate with the camera and if it is greater than or equal to 30, it would be flagged suggesting user is stressed, anxious or in an unsafe situation. Therefore, we also wanted to enhance the Presage model so that if a sudden phone movement is detected, such as if the phone is dropped, it would cause the AI agent to call the police in case of emergency.
Built With
- antigravity
- css
- elevenlabs
- gemini
- github
- html
- postgresql
- presage
- typescript
Log in or sign up for Devpost to join the conversation.