Below is my Second Brain info. Just copied what you provided. Going to change it as I get clearer on what I want. # Second Brain A personal system that knows what I know and what I'm doing. It ingests documents and connects to my calendar, email, and notes, then answers questions across all of it with citations back to the source — and eventually briefs me on my day. ## Status Greenfield — no code yet. This file exists to anchor the vision before implementation starts. ## Core idea One system, several sources of truth: - **Inbox folder** — documents dropped in for ingestion (PDFs, notes, articles, etc.) - **Calendar** — what's scheduled - **Email** — what's being communicated - **Notes** — what I've already written down Every answer the system gives should be traceable back to the specific source it came from (a document, an email, an event, a note) — no unsourced claims. ## Planned capabilities (roughly in order) 1. **Ingest** — watch/read the inbox folder, extract and index document content. 2. **Query with citations** — answer questions across ingested content, always pointing back to the originating source. 3. **Connect calendar, email, notes** — bring these in as additional queryable sources alongside documents. 4. **Daily brief** — proactively summarize the day ahead (schedule, relevant emails, open threads, notes) at a chosen time. ## Design principles - **Citations are non-negotiable.** Every synthesized answer must link back to source material. If a claim can't be traced, it shouldn't be stated as fact. - **Read-only against connected accounts** unless a feature explicitly requires write access (e.g., eventually drafting replies) — and that's opt-in, not default. - **Local-first where possible.** This holds personal data (documents, email, calendar); prefer local storage/indexing over sending everything to third-party services. - **Start with the inbox.** Get ingestion + citation-backed Q&A solid on static documents before wiring up live accounts — it's the smallest slice that proves the core loop.