Alpha · 0.x

Shipping now

What ships in the alpha today. The Alpha backlog has been groomed into an actionable delivery queue of 17 ready issues; blocked and human-only work is separated.

The app navigation groups it as Build, Monitor, Improve, Configure, Admin, and System.

Governed pipeline builder

Compose agents, schemas, connectors, and gates into a pipeline. Start from a library template or from scratch.

Human approval gates

Pause a run where it matters. One claim, one decision, an audit record either way.

Runs with evidence

Every run records a definition snapshot, outputs, timing, and error detail, so any run can be inspected after the fact.

Append-only audit trail

Every state-changing action produces a chained audit event. The history you inspect is what happened.

Evals and variants

Score agent outputs, raise eval proposals from real runs, and A/B test models.

Guardrails and cost controls

Schema checks and regex gates at the boundaries, spend limits and cost tracking on every run, a spend circuit breaker per pipeline, and an org-wide pause.

Cron and polling triggers

Scheduled and polled runs without external schedulers.

Monitoring

Error dashboard, notification log, and analytics across your pipelines.

Team administration

Users, teams, organization, SSO, and run environments.

Native installers

Install on Linux with the installer script, or on macOS from a signed and notarized .dmg, with Docker Compose and Helm alongside.

Self-hosted administration

License, email, run retention, and housekeeping on your own infrastructure.

Beta

Planned

The beta is the enterprise-readiness milestone. It does not add new product surface area; it proves the product is safe to install, operate, recover, upgrade, secure, and validate. Each theme is an outcome-level acceptance criterion, not a feature.

Enterprise deployment & high-availability failover

Supported production topologies, redundant app/worker operation, health checks, failover behaviour, and rehearsed failure injection with documented RTO/RPO targets.

Disaster recovery & restore rehearsal

Automated clean-environment and damaged-database restore rehearsals with integrity checks, point-in-time limits, and published restore evidence.

Upgrade, failed-upgrade & rollback proof

Clean install, historical upgrades, interrupted/failed migrations, application rollback, database restore plus image rollback, and post-upgrade verification across supported targets.

Capability validation matrix & release evidence

Inventory of every customer-visible capability mapped to real validation evidence: API contract, integration, browser E2E, failure injection, security/tenancy, upgrade and restore validation.

Critical journey & integration conformance suite

Risk-based E2E coverage for install, SSO/JIT/offboarding, RBAC, connectors, pipelines, triggers, HITL, recovery, audit export, and representative GitHub/GitLab plus issue-tracker flows.

Capacity, performance & operational limits

Supported limits for users, pipelines, triggers, queues, workers, audit volume, and database size, backed by load, soak, burst, and degraded-dependency tests.

Security assurance & customer trust pack

Threat model, data-flow/retention statement, secrets and encryption model, SSRF/egress controls, tenant isolation, SBOM, signed releases, vulnerability policy, and reusable security questionnaire.

Operator observability, support bundle & incident runbooks

Health model and alerts, actionable logs/metrics/traces, redacted support bundle, credential rotation, queue/stuck-run recovery, audit-chain handling, and rehearsed runbooks.

Supported deployment, install & procurement readiness

Official support matrix, clean-host installation, resource requirements, database/cache choices, licensing/offline activation, air-gap constraints, upgrade policy, and buyer-facing architecture diagram.

GA

Stability and ecosystem

General availability focuses on stability, the public API contract, and the self-hosted experience. These items set expectations for what we are working toward, but are liable to change.

Assistant - An assistant that can act across your workspace, configured and audited like every other pipeline node.
Stable public API contract - Semantic versioning with deprecation windows.
Compliance exports - Package the audit trail as evidence for your own audits.
Install and upgrade story - A hardened install and upgrade path across the native installers, Docker Compose and Helm, with a documented upgrade process.
Scale and high availability - Production deployment topologies.
Complete documentation - Every feature documented and onboarding polished.