← Back to home
Public roadmap
What has shipped and what we are building next. Updated regularly. Not a promise — a best-effort snapshot.
2026
shipped- Goals as the main screen — metric and project goals, milestones, the work under each goal in three columns, and a 7-day “What changed” digest
- Contributions — a goal’s metric moves only from recorded contributions, with a source (experiment or solution) and a maturity; recorded from the UI or by an agent
- Work that serves no goal — orphans listed on the main screen, plus coverage of work linked to goals
- Opportunity Sandbox and per-team Solution Boards — kanban and table views, drawers, archive and restore, stage transitions with scoring gates (soft warnings and hard blocks)
- Reusable scoring library — RICC, ICE, WSJF, and custom formula models; any gate can reference any model
- Import debrief — paste or upload a backlog, plan or research (PDF, DOCX, TXT, MD, CSV); opportunities grouped by goal, likely duplicates, work that serves no goal, draft goals for an empty workspace
- Dashboard — funnel, stage bottlenecks, gate health, driver load per team; activity feed and comments
- Workspaces — several per person, invitations by email and link, owner/admin/editor/viewer roles, per-team roles, audit trail
- MCP server — 53 tools for AI agents: goals, opportunities, solutions, experiments and their results, links, members, comments, scoring models, stages and gates; a Claude connector per workspace, personal access tokens and ready agent instructions in Settings
- Public showcase — publish a read-only view of your goals home at a secret link
- Security and compliance baseline — DPA, Privacy Policy, Sub-processors, Security page; EU data residency; row-level security as defence in depth
Q4 2026
next- Team load heatmap — who is drowning and who is idling, surfaced in the dashboard
Q1 2027
planned- Self-serve billing for Org
- One-click MCP connection (OAuth) instead of personal tokens
- Jira two-way integration — link Sorby opportunities to Jira epics; sync status both ways
- Linear native integration
- Hypothesis similarity search — find past experiments with similar hypotheses before you start