Inspiration
As people that prioritize privacy in the ever growing digital space, I wanted to have a way to use AI from a clean, user friendly way with all of my tools that I need day to day incorporated by default.
What it does
It onboards the user so the data they choose to share is stored 'in house' and then references it when needed to complete actions as dictated by the users requests. It uses external API's to get data into its local storage if it needs it to cache data. Then it can re-reference it from there if other users with the same requests or similar demographics request it.
How we built it
It is a phoenix web app built with the ash framework at its core. React was the best choice of front end for its rich libraries and library support. We use a ash codegen to create the types and rpc client, this lets us have complex frontend state with simple backend verification.
Challenges we ran into
Some of the API's we wanted to incorporate did not work well with our setup. We worked around this by setting up a cron like task to fetch the data to our database then work with our database instead of the raw api data/shape.
Accomplishments that we're proud of
Our web app properly functions and ultimately we have an application that can be served via the elixir bandit package and can feasibly be self hosted for an individual level of setup.
What we learned
We learned a lot about contributing to a project with more than just a single developer as we had to 'multi-thread' our efforts in order to work well together.
What's next for Issyoue
We plan to work out the bugs in our system for edge cases and make the whole project a lot easier to work on with a refactor.
Built With
- elixir
- ollama
- react
- typescript
Log in or sign up for Devpost to join the conversation.