Recover the why.
Return to an old project and recover the decisions, people, files, and evidence that shaped it, without rebuilding the story by hand.
A decision from an old project. A photo you can almost picture. A person you recognize but cannot place. Your digital life is full of answers buried across files, photos, projects, and relationships. Quill brings them back, with the source attached.
and shows you the evidence
Personal life search
No filename · no exact phrase · no problem
Search tools expect you to remember where something lives and what it was called. Human memory does not work that way. Quill starts with the fragments you still have and reconnects the rest.
Return to an old project and recover the decisions, people, files, and evidence that shaped it, without rebuilding the story by hand.
Describe the photo, place, event, or fragment you remember. Quill follows the clues even when you cannot name the file or date.
See how a person, project, decision, and artifact connect so a familiar name becomes useful context again.
Hybrid retrieval
Exact · semantic · graph
Quill runs multiple retrieval paths together, fuses their signals, and reranks the strongest evidence. Each method covers the blind spots of the others.
Fast keyword retrieval catches names, phrases, filenames, and the details you remember precisely.
Embeddings recover relevant material even when your question and the source use different language.
Entity grounding and bounded graph traversal reveal how people, projects, places, and artifacts relate.
Durable orchestration
Continuous · resumable · provenance-aware
Long-running workflows scan, extract, embed, ground, and connect the archive in bounded steps. Interrupted work resumes safely, and unchanged material is not needlessly processed again.
Notice what changed without rereading everything.
Turn varied personal records into useful text and metadata.
Place documents and entities in one shared semantic space.
Resolve identity and connect evidence to the knowledge graph.
Run exact, semantic, and graph search together.
Return the answer with a path back to the source.
Efficient intelligence
The system uses deterministic processing for facts it can know directly and model inference where judgment improves the result. One embedding can support multiple retrieval paths; only the strongest evidence earns more expensive reranking.
Independent search paths run together, then converge into one ranked result.
Exact matches, semantic similarity, graph context, and reranking reinforce one another.
Every useful answer preserves its origin and a direct path back to the supporting source.
The archive remains access-controlled and locally owned, with read-only source connections and explicit boundaries.
Quill · Coming soon
Quill is being developed by Palmetto Interactive and will be offered through membl. We are preparing a private preview for people who want their digital life to become useful, connected, and searchable without surrendering control of it.
See Quill on membl →