Inspiration
I personally missed the opportunity to ask my mom and grandparents about their childhood, important life milestones, and sweetest memories. And I am not alone. A lot of people missed this opportunity, because we never know when our loved ones will go away forever.
What it does
Legacy AI is an autonomous agent that acts as a living, self-updating digital extension of the user.
- Continual Ingestion: It monitors designated data streams (e.g., Slack, Notion, GitHub commits, Voice Memos) in real-time.
- Dynamic Adaptation: Instead of a static RAG knowledge base, it employs continual learning strategies to update its mental model without suffering from catastrophic forgetting. It distinguishes between ephemeral updates (e.g., "meeting at 5 PM") and core personality changes (e.g., "new coding style").
- Autonomous Action: It can proactively draft replies, summarize missed context, and answer questions as the user, reflecting their most current state of mind—not who they were six months ago.
How we built it
Using tools from the sponsor:)
Basically, the user is communicating through voice with the agent. And the agent, very patient, warm, and welcoming, is asking the questions, which allows us to uncover the stories and memories. As a result, Agent can create a book of memories, a family tree,
Challenges we ran into
The biggest challenge - create a specific set of ethical rules of communication with seniors, because it very sensitive target audience.
Accomplishments that we're proud of
Our biggest accomplishments are ahead:)
What we learned
Building a product is not easy, but very interesting
What's next for Legacy AI
Finish MVP and start to implement with the target audience.
Log in or sign up for Devpost to join the conversation.