Planning State — TFM aInventory V2 Reset

Date: 2026-04-22
Reset Type: Lost priority focus / refocus on v2
Scope Preserved: Mission, tech stack, validated features, constraints

What Changed

Why Reset Was Needed

Current phase (3) complete; unclear what v2 should prioritize
Too many potential features; no clear direction
Field deployments active but feedback not systematized

What We Kept

✓ Core mission: inventory + scanning + AI extraction + offline sync
✓ Tech stack: FastAPI, SQLite, Next.js (no changes planned)
✓ v1.14.6 work: Image adjustment modal, rotation-only simplification, zoom fix
✓ Constraints: LDAP auth, offline-first, multi-AI support

What We Added

Clear v2 feature scope (must-have, should-have, nice-to-have)
Field validation phase (phase 4) before feature development
Structured roadmap: Phases 4-7 with milestones
Success criteria for each phase

Current State

Version: v1.14.6 stable
Branch: dev
Last Work: Session 33 (ImageAdjustmentModal rotation-only, zoom fixes)
Deployments: 1-2 pilot sites using v1.14.6

Next Steps

Phase 4 Planning (/gsd-plan-phase 4):
- Deploy v1.14.6 to pilot sites
- Collect field feedback
- Fix mobile UX blockers
- Update deployment docs
Phase 5 Planning (after field validation):
- Implement batch operations
- Add search/filtering
- Add export/reports
- Based on field user feedback
Ongoing:
- Monitor field deployments
- Track issues/feedback
- Adjust roadmap based on real user needs

Key Assumptions

Field users are actively using v1.14.6
LDAP + offline-first are non-negotiable requirements
SQLite is sufficient for v2 scope
AI providers (Gemini + Claude) remain stable and affordable
No multi-tenant or multi-warehouse needs in v2

Decision Log

Date	Decision	Rationale	Status
2026-04-22	Reset planning to phases 4-7	Lost focus on v2 direction; field feedback needed	Active
2026-04-22	Phase 4: Field validation first	Gather real user needs before building v2 features	Planned
2026-04-22	Defer cropping, analytics, federation	Not needed for v2; revisit in v3	Out of scope

Risk Register

Risk	Impact	Mitigation
Field users have different priorities than assumed	High	Phase 4 validation; adjust roadmap based on feedback
SQLite hits scale limits with 10K+ items	Medium	Phase 6 scale testing; optimize queries if needed
AI provider costs become prohibitive	Medium	Monitor usage; have fallback provider (Claude if Gemini fails)
LDAP integration blocks new deployments	Medium	Phase 4 docs; simplify setup; provide troubleshooting guide

Last updated: 2026-04-22 after reset

2.8 KiB Raw Blame History