Roadmap
What's next
A transparent look at where ROW Recall is going.
Staged Execution Strategy
ReCall ships in stages — every stage proves one capability before the next is built. The north-star metric is memory-assisted AI sessions.
Stage 0 — Foundation (0.5.x) · mostly done
- AES-256 / SQLCipher encryption, secure key management, machine-bound licensing, cross-device recovery.
- Remaining: final signed build (blocked on GitHub billing), licensing account/device rework, key lifecycle + threat-model tests.
Stage 1 — MCP Proof (0.6.0) · DONE (2026-08-13)
- recall_context() — the single killer MCP tool — plus recall_search; minimal Context / Search / Action surface.
- A/B benchmark harness for the measured claim.
Stage 2 — Context Intelligence (0.7.0) · DONE (2026-08-13)
- Persistent memory layer (project / decision / failure / agent memories) with provenance, confidence and ranking.
- Workspace Intelligence day-1 card — context within minutes of adding a workspace, no history required.
Stage 3 — Memory Flywheel (0.8.x) · next
- Agent & experiment memory, provenance/confidence lifecycle, and Ask ROW — a natural-language window into your timeline.
Stage 4 — Recovery Intelligence (0.9.x)
- Last-known-good detection, state comparison and safe restore.
Stage 5 — Team (1.0.x)
- Shared project memory, agent handoffs, permissions — 1.0 means team-ready.
Stage 6 — Enterprise + Mobile
- Governance, SSO, self-host, mobile control center — adoption-gated.