Field notes · Product preview
You Remember the Thing. Not Where It Lives. Meet Quill.
You know the information exists. You remember enough to know it matters. What you do not remember is the filename, the folder, the app, the exact wording, or the year.
It might be the reason a project changed direction. A photo you can almost picture. A person whose name is familiar but whose connection has gone fuzzy. A promise made in a document you have not opened in years.
Ordinary search asks you to remember how information was stored. Human memory does not work that way. We remember fragments: what something meant, who was involved, what happened before it, and why it mattered.
That is the problem Quill is built to solve.
Personal life search
Quill is a coming-soon personal search product from membl. It turns a private archive of files, photos, projects, and relationships into one connected place to ask questions.
The useful version of this idea is not another folder browser or a chat box sitting on top of a vector database. It should help with the moments where memory is incomplete:
- Recover the why. Return to an old project and reconstruct the decisions, people, and evidence that shaped it.
- Find the moment. Describe a photo, place, event, or fragment even when you cannot name the file or date.
- Restore the context. See how a person, project, decision, and artifact connect so a familiar name becomes useful again.
- Resume the work. Re-enter something you built months or years ago without rebuilding its history by hand.
The promise is simple: ask the way you remember it, then get the trail back.
Search by words, meaning, and connection
The technology matters only because it helps solve that problem well.
Quill combines exact retrieval, semantic search, and graph relationships. Exact search finds the names, phrases, and filenames you still remember. Semantic retrieval finds relevant material even when your question uses different language. Graph retrieval follows bounded connections among people, projects, places, and artifacts.
Those paths run together. Quill combines and reranks their strongest evidence before producing an answer. The result preserves a path back to the source, so the system can show why an answer surfaced instead of asking you to trust an unsupported summary.
Behind the search, durable workflows keep the archive current. They notice changes, extract useful text and metadata, create embeddings, connect entities, and resume safely when work is interrupted. Selective inference keeps deeper model work focused on the evidence where it improves the result.
The complexity stays underneath. The search should simply work.
Coming soon from membl
Quill will be offered through membl alongside Sweetgrass, Lantern, and Marsh. It is still in development and is not publicly available today. The first release is being shaped around private, access-controlled search with provenance built in from the beginning.