Merge pull request 'Plaud comparison + feature-gap backlog' (#4) from feature/plaud-review into main

Reviews the Plaud line (Note, Note Pro, NotePin/S) to make sure the OpenScribe
roadmap covers expected capabilities, and records where OpenScribe deliberately
differs (bring-your-own-AI, self-hosted, open).

What changed:
- docs/plaud-comparison.md: feature map (Plaud vs OpenScribe status), a derived
  backlog, and what we deliberately do not copy. Sourced from Plaud's site (July 2026).
- state/TODO.md: adds a Plaud-derived backlog section (diarisation, summary templating,
  Ask-AI/RAG, custom vocabulary, multimodal, mark-a-moment, auto-title, tags/folders/
  search, share links, optional OLED display, calendar) and a pluggable-AI-providers
  milestone.

Why:
- The user asked to review Plaud for other todos; this captures them and frames the
  bring-your-own-AI differentiator that the next feature implements.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

docs/plaud-comparison.md

@@ -0,0 +1,71 @@
+<!-- SPDX-License-Identifier: CC-BY-SA-4.0 -->
+# Plaud comparison and feature gap analysis
+
+A review of the Plaud product line (Note, Note Pro, NotePin / NotePin S) to make sure the
+OpenScribe roadmap covers what people expect from this class of device, and to be clear
+about where OpenScribe deliberately differs. Reviewed July 2026 from Plaud's site and
+listings (see Sources).
+
+OpenScribe's core difference: **you bring the AI.** Plaud routes everything through
+Plaud's own cloud AI on a subscription. OpenScribe lets the owner point transcription and
+summarisation at any open-standard endpoint (OpenAI-compatible / local) or a commercial API
+(OpenAI, Anthropic, Gemini), self-hosted or not, with no subscription and no lock-in.
+
+## Feature map
+
+| Capability | Plaud | OpenScribe status |
+|-----------|-------|-------------------|
+| One-button voice recording | Yes | Done (M1) |
+| Long recordings to storage | Yes (cloud) | Done (M1, microSD) |
+| Sync to phone app | Yes (BLE/WiFi) | M2 (WiFi) done; M4 (BLE); M7 (app) |
+| AI transcription, many languages | 112 languages | M5 (provider-driven; language set by provider) |
+| Speaker labels / diarisation | Yes | Backlog (added below) |
+| Custom vocabulary | Yes | Backlog |
+| AI summaries | Yes | M6 |
+| Summary templates (meeting, to-do, mind map, custom) | 10,000+ | Backlog (templating) |
+| Multidimensional / role-based summaries | Yes | Backlog |
+| Ask AI / chat grounded in the recording | "Ask Plaud" | Backlog (RAG over transcript) |
+| Multimodal: attach text/images to a recording | Yes | Backlog |
+| Highlight / mark key moments while recording | Yes | Backlog (device long-press -> timestamp) |
+| Auto title + auto formatting | Yes | Backlog |
+| Tags / folders / search | Yes | Backlog (server + app) |
+| Calendar integration | Yes | Backlog (nice-to-have) |
+| Share summaries / export | Yes | Export done in API (M5); share links backlog |
+| On-device status display (InstantView) | Note Pro | Hardware future (optional OLED variant) |
+| Phone-call recording | Yes (MagSafe) | Out of scope (no phone-call tap) |
+| Choice of AI provider (open or commercial) | No (Plaud cloud only) | **OpenScribe differentiator** |
+| Self-hostable, no subscription | No | **OpenScribe differentiator** |
+| Open source, open API, open hardware | No | **OpenScribe differentiator** |
+
+## New backlog items derived from this review
+
+These are added to `state/TODO.md` (Plaud-derived backlog):
+
+1. Speaker diarisation + labels (server), with rename-speaker in the app.
+2. Summary templating: built-in templates (meeting notes, action items, to-do list, mind
+   map, decisions) + user-defined templates; role-based multidimensional summaries.
+3. Ask AI: chat grounded in a recording's transcript (RAG), reusing the chosen LLM provider.
+4. Custom vocabulary / prompt hints passed to the transcription + summary providers.
+5. Multimodal attachments: text notes and images attached to a recording.
+6. Mark-a-moment: device long-press during recording writes a highlight timestamp;
+   surfaced in the app and used to focus summaries.
+7. Auto title + auto formatting of transcripts and summaries.
+8. Library organisation: tags, folders, full-text search (server + app).
+9. Share links / export of summaries (extends the existing export endpoint).
+10. Optional hardware variant with a small OLED status display (InstantView equivalent).
+11. Calendar integration (nice-to-have) to attach recordings to events.
+
+## What we deliberately do not copy
+
+- Phone-call recording via a phone accessory (Plaud's MagSafe call capture). Out of scope.
+- Subscription tiers and per-minute transcription quotas. OpenScribe has none by design; the
+  cost is whatever your chosen AI provider charges (zero if self-hosted).
+- A closed cloud. Everything stays on infrastructure the owner controls.
+
+## Sources
+
+- [Plaud NotePin](https://www.plaud.ai/products/plaud-notepin)
+- [Plaud Note Pro](https://www.plaud.ai/products/plaud-note-pro)
+- [Plaud Note](https://www.plaud.ai/products/plaud-note-ai-voice-recorder)
+- [Plaud NotePin S](https://www.plaud.ai/products/plaud-notepin-s)
+- [Plaud: AI Note Taker (App Store)](https://apps.apple.com/us/app/plaud-ai-note-taker/id6450364080)

state/TODO.md

@@ -27,6 +27,30 @@
 - [ ] M9 Hardening + OTA: API auth, credential handling, OTA update, docs handoff.
       (branch `feature/hardening-ota`)
 
+## Pending - AI provider flexibility
+
+- [~] Pluggable AI providers (server): transcription + LLM can target any open-standard
+      endpoint (OpenAI-compatible / local faster-whisper / Ollama) OR a commercial API
+      (OpenAI, Anthropic, Gemini). Config-driven, no lock-in. (branch
+      `feature/server-ai-providers`) - provider layer + HTTP providers landing now;
+      local faster-whisper wiring completes in M5.
+
+## Pending - Plaud-derived backlog (see docs/plaud-comparison.md)
+
+- [ ] Speaker diarisation + labels (server); rename-speaker in the app.
+- [ ] Summary templating: built-in templates (meeting/action-items/to-do/mind-map/
+      decisions) + user templates; role-based multidimensional summaries.
+- [ ] Ask AI: chat grounded in a recording's transcript (RAG), via the chosen LLM provider.
+- [ ] Custom vocabulary / prompt hints for transcription + summary.
+- [ ] Multimodal attachments: text notes + images attached to a recording.
+- [ ] Mark-a-moment: device long-press during recording -> highlight timestamp, surfaced
+      in the app and used to focus summaries.
+- [ ] Auto title + auto formatting of transcripts and summaries.
+- [ ] Library organisation: tags, folders, full-text search (server + app).
+- [ ] Share links for summaries (extends the export endpoint).
+- [ ] Optional hardware variant with a small OLED status display (InstantView equivalent).
+- [ ] Calendar integration (nice-to-have).
+
 ## Done
 
 - [x] Dev testing + emulation harness - native unit tests (wav/range), Wokwi emulator,