Inspiration
Most of my experience has been in full-stack development, and I hadn't had the chance to explore the AI side of computer science. This was my opportunity to dive deep into it, and Photon's track really interested me: pairing AI with iMessage makes talking to an agent easy and accessible. I wanted to build something I'd genuinely use, and this project gave me a real reason to finally commit to Google Calendar.
What it does
Beacon takes the stress out of forgetting important events by managing your Google Calendar over iMessage. You can: • View, create, edit, and delete events just by texting • Get reminders up to 24 hours before upcoming events, plus an optional daily briefing • Invite friends to events by email, right from iMessage • Find out when to leave to arrive on time, based on how you travel (driving, transit, walking, or biking) • Find places nearby and have Beacon remember your preferences
How we built it
Through spec-driven AI development I was able to leverage Cursor to design, plan, implement, test, and hand knowledge to next sessions. Created an idea with a vision in how it flows, how interactions are handled, and high level decision making in the architecture guiding Cursor at every step of the way.
Challenges we ran into
Lots of different setup from grabbing Cursor CLI, creating accounts on services like xAI, Cursor, Figma, Photon, and Hermes. Decision making when it came to the natural push and pulls of feature development thinking of trade offs like getting rid of prospect of group chats in favor of not leaking personal data (if someone use the calendar tool to look at calendar very likely every in the group chat could see their calendar).
Accomplishments that we're proud of
Creation of a viable product that does what it is advertised to do, learn about AI tooling and understanding how AI interacts with the tools it is given, memory achieved through SQLite storing of potential important pieces of information.
What we learned
Lots about how to use API keys (following best practices to not leak sensitive data), how to utilize Markdown files in order to give context to AI agents allowing them to understand architecture, coding standards, and previously determined data. Just how powerful AI can be!
What's next for Beacon
The sky is far from the limit. Beacon can continue on in many areas looking towards integrating with the iOS native environment (Alarms, Calendar, and many many other apps on the phone). Looking towards a safe group chat bot that would not leak personal data but allow for events to be created jointly. Control of any app that has API endpoints to make tools for Beacon to interact with its environment.
Log in or sign up for Devpost to join the conversation.