Reachly

Inspiration

On the drive here from Kansas, we talked about how staying connected has become strangely exhausting these days, whether that’s remembering birthdays, replying to messages, or just reaching out. Even with endless social platforms, real connection is slipping away. This conversation inspired our idea to build something that bridges those quiet gaps in our relationships with simple but impactful touchpoints.

What it does

Sends reminders to contact individuals that you haven’t contacted to in a while
Provides general automated text message suggestions to start a conversation
Analyzes previous conversation to create context-adapted conversation starters
Adds key events & dates regarding different individuals, such as birthdays, onto the calendar
Schedules in-person/virtual meeting dates based on availability and adds date to calendar
Analyzes frequency/pattern of contact and reminds the user if the pattern is disrupted

How we built it

We built an application using Flutter and displayed an iPhone view in Windows. Since none of us had a Mac to sideload it onto our iPhones, we used NVIDIA's API key to connect the model to our Python Flask server, which then linked to our Flutter service. Moreover, we used 8 AI agents and created a systematic hierarchical agent workflow.

Challenges we ran into

We initially tried to deploy our NVIDIA nemotron model with Vultur but faced plenty of credit issues for getting the cloud GPU, so we decided to stick with the API interface and Brev server provided by NVIDIA itself.

Accomplishments that we're proud of

We are proud of achieving an effective and smooth connection between our NVIDIA model with its multi-agent workflows to Flutter and our Python backend.

What we learned

We learned how complex and intricate running a high-performance model of Nvidia on Flutter is and its agentic abilities, which exceed a simple chatbot from an API call. Furthermore, we learned to improve prompt engineering and debugging code.

What's next for Reachly

Looking ahead, we plan to integrate a gamified rewards system using blockchain technology to incentivize users. Similar to Snapchat’s streaks, users will earn tokens or coins for consistently engaging with the app, whether it’s reaching out to friends, scheduling catch-ups, or staying on top of important dates. This rewards system will not only keep users motivated but also help improve their social habits over time, fostering deeper connections and encouraging frequent use of the app.