Personal AI Voice Assistant

Here's a draft for your Devpost submission, tailored to the CockroachDB × AWS "Agentic Memory" hackathon theme. I've made reasonable assumptions about your tech stack — swap in specifics where you have them.

Inspiration

Voice assistants today forget everything the moment you close the app. We wanted to build something different — an assistant that actually remembers who you are, what you've asked before, and how you like things done, so every conversation builds on the last instead of starting from zero.

What it does

Personal AI Voice Assistant lets you talk naturally to complete tasks, ask questions, and get help — while remembering context across sessions. It recalls your preferences, past requests, and ongoing tasks, so you don't have to repeat yourself every time you open the app.

How we built it

We used AWS for voice processing and AI inference, paired with CockroachDB to store and retrieve persistent memory — user preferences, conversation history, and task context — in a scalable, distributed way. This let us build an agent that reasons over both the current conversation and everything it has learned about the user over time.

Challenges we ran into

Balancing memory recall with response speed was tricky — pulling in too much context slowed things down, while too little made the assistant feel forgetful. We also had to design a memory schema that could scale without becoming unwieldy as conversation history grew.

Accomplishments that we're proud of

Getting the assistant to genuinely feel "aware" — referencing something from a previous session naturally, rather than needing to be re-told. We're also proud of how smoothly the voice interface handles real-time interaction while querying memory in the background.

What we learned

We learned a lot about designing memory architectures for AI agents — deciding what's worth remembering, how to structure it, and how to retrieve it efficiently. We also got hands-on experience with AWS's voice and AI tooling alongside CockroachDB's distributed SQL capabilities.

What's next for Personal AI Voice Assistant

Next, we want to add multi-device sync, smarter memory summarization (so old context doesn't bloat the database), and proactive suggestions — where the assistant anticipates needs based on patterns it's learned, instead of only responding to direct requests