Two commercial targets shape every decision: the data governance execution layer for regulated enterprises (plan-time tag and policy enforcement as typed predicates), and displacing Oracle Exadata in regulated-industry customers via an Iceberg lakehouse and federation across the existing data estate.
The editorial documents. Slow-read, authoritative. Open one when you need to align on what we're building and why; not when you need to move fast on a ticket.
Commercial thesis, phase plan, exit criteria, hard constraints. The "why" behind every roadmap call. Last refresh: April 2026.
→Seventeen sections of technical commitments — catalog, identity, governance, distributed execution, lineage. The "what does the engine commit to."
→The data governance execution layer thesis. How Peaka explains itself to a regulated-industry CIO in three sentences and one diagram.
→Where Spice AI overlaps, where it doesn't, and what the v3.0 strategy means for the competitive frame.
The dashboard surfaces. Scannable, current. Open one when you're picking up a ticket, reviewing a PR, or reporting status.
How strategy, specs, and code stay in sync. The seven-step loop, CI gates, AI reviewers, branch protection.
→Phases 0 → 5 with status, exit criteria, and current progress. Timeline view at the top.
→Mirrored from CLAUDE.md. Every change must satisfy all of them. Plus the "what NOT to do" list.
What's in flight: four Phase 0.5 tasks, plus the next Phase 1 epics as they open. Each links its Linear ticket and PR.
Where the actual work happens. This dashboard summarizes; these are authoritative.