Roadmap
High-level direction for where DriftQ is heading next.
Note: this roadmap is directional, intentionally high-level, and subject to change.
DriftQ is the overall project/org. DriftQ-Core is the main GitHub repo that contains the durable broker and replayable runtime foundation.
At a glance
v1: DriftQ-Core (MVP reliability foundation)Completed
v1 is the reliability layer: the broker semantics you need before any workflow system can be trusted.
- Leasing - safe ownership and crash recovery
- Retries - backoff and max attempts
- DLQ - strict dead-letter routing
- Idempotency - duplicate protection under redelivery
- Observability - inflight, lag, DLQ totals, and backpressure visibility
v2: Replayable Workflow RuntimeCompleted
v2 turns the reliability foundation into a practical runtime for replayable AI workflows, safe retries, and time-travel debugging.
- Replayable runs - re-drive from the step that changed instead of rerunning everything
- Workflow primitives - durable execution patterns for real production workflows
- Budgets and throttles - guardrails around retries, timing, and concurrency
- Developer tooling - CLI and debug flows for inspecting, replaying, and validating runs
- Operational visibility - better insight into timing, run state, and runtime behavior
v3: Multi-Agent Runtime and Real-Time AICompleted
v3 turns DriftQ into a governed runtime layer for multi-agent systems, secure tool execution, real-time AI workflows, and production-grade debugging.
- Multi-agent runtime - coordination patterns, agent state, semantic memory, and Brain v1 routing
- Trust and governance - RBAC, tenant isolation, policy rules, risk checks, and runtime approvals
- Real-time and tools - low-latency topics, secure tool gateway, MCP-ready governance, and side-effect receipts
- Safety loops - evals, HITL, release engineering, self-healing recovery, and approval-before-commit
- Production readiness - OpenTelemetry, forensics, replay branches, workflow releases, and adaptive routing
Status: the v3 runtime foundation is complete in DriftQ-Core. The remaining SDK work is being split into separate repos.
v3.x: Maintenance and EcosystemIn progress
This is the current maintenance phase for DriftQ-Core while the broader product roadmap is being reevaluated.
- Runtime polish - refactoring, cleanup, and performance work across the core runtime
- SDK ecosystem - splitting SDK work into dedicated repos and improving developer ergonomics
v4: DriftQ CloudOn hold
v4 remains a potential future direction where the trust layer becomes operable as a managed platform for AI-first applications.
Status: this phase is not in active development right now, and public details will stay intentionally broad unless work resumes.
v5: Adaptive Intelligence LayerOn hold
v5 remains a longer-term exploratory direction: ML-assisted routing, failure intelligence, and dataset-driven optimization built on top of the evidence the runtime collects.
Status: this is not on the active build path today. If revisited later, it should stay evidence-driven, with heuristics and explainable routing coming first.