OpsDevAI
Shipping continuously

Changelog

Every release of OpsDevAI — new capabilities, improvements, and fixes. Newest first.

RSS coming soon · Follow on X for release notes.

  1. Predictive autoscaling, take three

    v1.4.0
    • NewToken-aware autoscaler now plans capacity 60 seconds ahead of demand using your last 14 days of traffic.
    • NewPer-workload SLO budgets — scale aggressively when budget allows, conservatively when it doesn't.
    • Improvedp99 cold-start latency on edge regions reduced from 412ms → 187ms.

  2. AI-screened CI/CD: blast-radius v2

    v1.3.2
    • NewRisk reports on every PR — blast radius, blast surface, blast confidence.
    • ImprovedAuto-rollback decision time cut to <800ms (was 2.4s).
    • FixedMerge queue no longer stalls when a downstream check is reverted mid-flight.

  3. Live sandbox CLI

    v1.3.0
    • Newopsdev sandbox — spin up an ephemeral slice of a real cluster against your branch, in-browser.
    • NewSandbox sessions auto-expire and never accept production traffic.
    • ImprovedCLI install size dropped from 38MB → 11MB.

  4. Hexagonal mesh telemetry

    v1.2.4
    • ImprovedPer-hop latency now visible in the console with sub-millisecond resolution.
    • ImprovedeBPF datapath observability — no sidecar, no measurable overhead.
    • FixedCross-region replication lag spike during regional failover correctly attributed.

  5. Zero-trust by default

    v1.2.0
    • SecuritySPIFFE-based workload identity now enabled by default on all new clusters.
    • SecuritymTLS rotation cadence configurable per workload; default reduced from 24h → 4h.
    • NewPer-tenant evidence export — SLSA provenance bundled into every release artifact.

  6. Self-healing clusters

    v1.1.0
    • NewNode-level drift detection — automatic remediation before a single request is dropped.
    • NewWorkload-aware eviction: evictions ranked by historical impact, not pod-creation order.
    • ImprovedMTTR for upstream provider outages reduced 4.2× across pilot accounts.

  7. OpsDevAI Core, generally available

    v1.0.0
    • NewAutonomous ML orchestration engine — learns your workloads and pre-warms capacity before spikes.
    • New120+ edge regions stitched by anycast + BGP-aware routing.
    • NewConsole: cluster metrics, CI/CD pipelines, OpsDevAI configuration, infrastructure logs.

  8. Private beta opens

    v0.9.0-beta
    • NewFirst 12 design partners onboarded — combined fleet of ~3,400 nodes.
    • NewRead-only console + CLI shipped to all design partners.