Highlights
What this release added
- Closed Milestone 10 sprint documentation by replacing Sprint 45 and Sprint 46 planning artifacts with canonical completion records, including a deferred closeout for mobile scope and a final connector closeout record for provider rollout and webhook audit delivery.
- Added an April 2026 AI competitiveness audit, rebaseline plan, and active AI backlog to recenter Wizzo around the main chat, grounded tools, integrations, and deliberate de-scoping of stale pilot-era AI surfaces.
- Graduated connector positioning for the 2026 AI reset: Google Drive, Gmail, and Calendar are now treated as core work connectors, while GitHub remains supported as an advanced technical connector.
- Improved AI memory recall so chat context now prioritizes execution-oriented memories such as plans, follow-ups, blockers, quest context, deadlines, and recent progress over generic preference replay.
- Completed the agentic planning decision: the legacy planning flow is now framed as an internal Labs fallback, with neutral deep-research telemetry and a deprecated compatibility route.
- Made main chat tools easier to discover by adding explicit composer actions for web search, deep research, JavaScript execution, image generation, and attachments.
- Completed the voice differentiation pass by keeping OpenAI-managed voice I/O as the product path and freezing bespoke voice-analysis ML behind an explicit legacy flag.
- Completed the docs reset by moving active planning to `docs/backlog/`, treating sprint docs as archive-only, and replacing per-doc version stamps with backlog/current-as-of guidance.
- Cleaned up Labs positioning so gated pilots render from actual feature-flag state and Image to Quest states its experimental scope without linking to missing docs.
- Added an AI value-signals report command that ties AI usage to tool success, quest follow-up, and return-session evidence before expanding the roadmap.
- Consolidated testing docs by moving sprint-specific guides and reports into sprint history, leaving active testing docs for standards and reusable validation guidance.
- Added user-facing account data export and confirmed deletion flows covering core product, AI, notification, integration, ML-derived, and community records.
Fixes
What this release corrected
- Updated milestone roadmap and backlog references so Milestone 10 no longer presents Sprint 46 as an active planning sprint after PRs #76, #77, and #78 merged.
- Refreshed canonical operator and product AI docs (`AGENTS.md`, product overview, AI handbook, active backlog) from stale `v0.42.0` / Sprint 42-43 framing to the current Public Beta runtime posture.
- Rebased the voice stack from ElevenLabs to OpenAI-managed audio models, replacing the legacy TTS endpoint, adding server-side OpenAI transcription, and normalizing old persona voice IDs to valid OpenAI voices.
Known issues
What remained constrained
- _None yet_