Most family stories and memories are passed down generation by generation in piecemeal and lost entirely to time.

Few of us will ever have a biographer to sit down with us for hours on end, ask us about our lives, and transcribe our answers. This is a problem that strikes close to home for our team as we have aging grandparents and relatives and are acutely aware of how much knowledge, tradition, and history we stand to lose if we don't act to preserve these memories soon.

What it does

Quilt is your personalized, automated family historian.

Quilt helps families

  • facilitate meaningful conversations through thoughtful prompts
  • automatically transcribe conversations and record highlights
  • automatically compose beautiful biographies and memoirs for your entire family
  • create and your family tree and securely explore your genealogy

What's next for Quilt

We started this entire project from scratch at 6pm on Friday. This was an idea we had been chewing around on for a while, and YC Hackathon was a great catalyst to take our first stab at fleshing it out. We plan to continue work on Quilt since it's a product we are building for our own families. We hope to use Quilt to make it easier to capture, organize, and share our family memories for posterity before it's too late.

In time, we plan to integrate deeper AI components into Quilt to make the conversations less structured (automatically infer topics, chronology, etc.) and explore deeper AI narrative technologies to craft personalized, high quality prose.

How we built this

Using state-of-the-art speech to text through Azure, we have built a platform that dynamically stores and understands your family history on a deep level. Using a graph datastructure, we are mapping family members to events, major life questions, and highlights of their lives. All data is collected through a live messaging interface using WebRTC, where participants simply lean back while their conversation is saved, transcribed, and mapped back to the individual person's timeline.

Share this project: