Claude
17c7f5033f
Decision (Q15 resolved): start with text-only via TEI + bge-m3 in Phase 3.1, prepare data shapes so Phase 3.2 (visual collection via infinity + jina-clip-v2) is a pure additive operation — no rename, no schema migration, no n8n rewiring. Concretely: - Qdrant collection renamed nexa_knowledge → nexa_knowledge_text (1024-dim for bge-m3) with modality-aware payload (modality, source_type, media_uri, graph_iri, content_hash, context). Visual placeholder schema committed alongside (qdrant_schema_visual.json, 768-dim, jina-clip-v2). - Image attachments captured in 3.1 are recorded in GraphDB as nexa:Note with nexa:modality "image" + nexa:pendingVisualIndex true; the 3.2 backfill workflow picks them up and embeds. No data lost between phases — the queue is the GraphDB itself. - RDF schema (docs/08) gains nexa:modality, nexa:mediaUri, nexa:vectorCollection, nexa:pendingVisualIndex from day one. - docs/02 roadmap split: 3.1 = text RAG (Path A), 3.2 = visual collection (Path C), 3.4 = Ontotext GraphDB. - docs/09 grows a "Phase add-on: visual collection (Phase 3.2)" section with the TEI→infinity swap, second collection create, LiteLLM second model registration, and the SPARQL-driven backfill query. - New open questions: Q16 (queue ergonomics + does SAIA already proxy an embed model?), Q17 (reuse Immich's CLIP for photo-library queries?). - docs/03 + CLAUDE.md updated so future runs use the new collection names and don't re-decide the staging.
2.6 KiB
2.6 KiB
02 — Roadmap & Phases
Iterative build-out, value-first. Each phase is shippable on its own.
Phase 1 — The Spine (connectivity)
Focus: datapath between Memos, n8n and SAIA. Milestone: Nexa replies on Memos and answers simple questions.
- 1.1 Memos ↔ n8n bridge — webhook ingress, content filter, comment-back.
- 1.2 SAIA integration — call LiteLLM proxy from n8n; chat via Memos comments.
- 1.3 Git-sync — periodic export of n8n workflows into this repo.
Phase 2 — Senses (input channels)
Focus: e-mail filter and the Work-vs-Personal router. Milestone: Nexa distinguishes work tasks from personal tasks without manual tags.
- 2.1 Email butler — filter & summarize the ~20 daily personal mails.
- 2.2 RSS morning digest — curated daily summary in Memos.
- 2.3 Bluesky antenna — store liked posts as knowledge snippets.
Phase 3 — Memory (Qdrant & RAG)
Focus: Qdrant + graph layer for retrieval-augmented answers. Milestone: RAG works — Nexa answers from archived notes; images captured today are queued for visual indexing later.
- 3.1 Text RAG (Path A) — TEI +
bge-m3, single Qdrant collectionnexa_knowledge_text.#nexa:askreads it before answering. Image attachments are recorded asnexa:NotewithpendingVisualIndexbut not embedded yet. - 3.2 Visual collection (Path C) — swap TEI → infinity, add
jina-clip-v2, createnexa_knowledge_visual, backfill all queued image notes. RAG workflow gets a parallel branch for visual hits. See 09 § "Phase add-on: visual collection". - 3.3 S3 long-term archive — Qdrant snapshots (both collections) to the S3 server.
- 3.4 Ontotext GraphDB — SPARQL pillar; structural queries combined with vector hits in
#nexa:ask. See 08.
Phase 4 — Motor (organization & action)
Focus: calendar integration and time-boxing. Milestone: Nexa proactively proposes morning focus slots in the work calendar.
- 4.1 Nextcloud Tasks sync — bidirectional status sync (Memos ↔ NC Tasks).
- 4.2 Time-boxing agent — calendar-aware slot suggestions.
- 4.3 Karakeep curation — automated review cycle for saved links.
Phase 5 — Daily integration (polish)
Focus: Home Assistant Voice and monitoring. Milestone: Nexa speaks via HA-Voice and surfaces critical system states proactively.
- 5.1 HA Voice — Wake-word integration, Wyoming protocol.
- 5.2 System monitoring — Proxmox / Backrest alerts into the Memos system feed via ntfy.