Changelog
Every update, improvement, and fix shipped to the VxCloud platform. Subscribe to stay informed.
Sales & Marketing Debuts — SalesShift
SalesShift launches at salesshift.vxcloud.io: a full sales and marketing workspace with a prospect pool, outbound sequences, a real calendar with emailed invitations, quotes and invoices, and in-app calls — sitting on the same tenant isolation as the rest of the platform.
- NewSalesShift is live at salesshift.vxcloud.io — CRM contacts, leads, deals and opportunities in one workspace
- NewProspect pool with an enrichment crawler: search, reveal and import leads, with role-address filtering built in
- NewOutbound sequences and campaigns with a real scheduler — step timing, per-run locks, and delivery tracking
- NewCalendar with genuine invitations: ICS attachments, RSVP handling, booking links, and reminders that actually send
- NewQuotes, e-signature links and enterprise invoice PDFs, emailed straight from the workspace
- NewIn-app audio and video calls with screen share, plus guest chat for CRM contacts who have no account
- NewSupport desk on vxcloud.io/support: open a ticket without an account, chat live, and get email at every step — opened, in progress, and closed
- ImprovedEvery ticket is also mirrored to the shared support inbox, so a request is never lost when nobody is at the keyboard
- SecurityMessaging rebuilt on a zero-trust tenant model: channels, groups, direct messages and search are scoped to your own organization, with outsiders admitted only by invitation
- SecurityTwo platform-wide channels only — #community, open to everyone, and #open-discussions, where you request access and an admin approves
- SecurityHardened the messaging plane: authenticated event stream, attachment path containment, scoped mention and call fan-out, and org-scoped saved messages and mentions
Agent Control — The Programming UI
Agent Control gains a real programming surface: write code against your agents, run it, and watch it execute. Backed by a live MCP client, dataset management, and a knowledge base that actually retrieves.
- NewProgramming UI in Agent Control — author, compile and run code against your agents from the console, with output streamed back
- NewMulti-language execution: Python, C++ and Go templates that compile and run on the node, not in a mock
- NewReal MCP client — connect Model Context Protocol servers and call their tools from an agent, with the tool catalog discovered live
- NewDatasets tab: upload JSONL, inspect rows, and attach a dataset to a fine-tune without leaving the page
- NewMemory Bank — agents keep durable memories across sessions, with extraction, consolidation and runtime injection
- NewRAG engine upgrades: knowledge-base ingestion into FAISS plus a cross-encoder / LLM reranker on the query path
- NewEphemeral model deploy — spin a model onto a TTL-bounded VM with inference and automatic HTTPS, then let it expire
- ImprovedCode interpreter, execution tracing, guardrails and prompt versioning brought in line with the major agent platforms
- ImprovedAgents and models deploy to your own worker VMs from the marketplace or vxcli, and are chattable the moment they are up
- FixedTalk-to-Model now falls back to a local runtime when the hosted endpoint is unavailable, instead of failing silently
VxNode Goes Public — Self-Hosting for Everyone
The vxnode repository is now public, and full self-hosting is here. Run the complete VxCloud node on your own VM with a single script — multi-arch images, automatic SSL, and fleet self-updates included.
- Newvxnode public repository released — deploy the VxCloud node engine on your own infrastructure
- NewOne-script self-hosting: tenant_setup.sh provisions the container, Nginx reverse proxy, and Let’s Encrypt SSL on any Ubuntu VM
- NewMulti-arch Docker images (linux/amd64 + linux/arm64) — runs on Graviton, Ampere, and standard x86 hosts out of the box
- Newvxcli installed automatically inside every node from the public download channel, alongside Claude Code, Codex, and Gemini CLIs
- ImprovedFleet self-update channel: self-hosted nodes poll the release manifest and upgrade themselves with health-gated rollback
- ImprovedHardened container defaults for self-hosted nodes — dropped capabilities, no-new-privileges, and resource limits auto-sized to the host
- SecuritySelf-hosted images ship with zero baked-in credentials — all secrets are runtime-mounted, never in the image or build context
VxCloud Mobile — Now Available
The native VxCloud mobile app is ready for deployment, bringing your entire control plane to iOS and Android with biometric login, push alerts, and on-the-go provisioning.
- NewNative VxCloud mobile app for iOS and Android with full dashboard, node management, and Cloud Shell access
- NewPush notifications for deployments, incidents, budget alerts, and approval requests with deep links into the console
- NewBiometric sign-in (Face ID / Touch ID / fingerprint) backed by hardware-secured token storage
- NewProvision, start, stop, and SSH into nodes directly from your phone — multi-cloud across AWS, GCP, Azure, and Alibaba
- ImprovedOffline-aware UI that queues actions and syncs automatically when connectivity returns
- ImprovedMobile-optimized log and metrics streaming with reduced data usage on cellular networks
- SecurityEnd-to-end encrypted session tokens with automatic revocation on device loss from the security dashboard
VxThinker — Public Release
VxThinker, our reasoning model for infrastructure and DevOps, is now released for public access on all plans — moving out of private preview with general availability across the platform.
- NewVxThinker reasoning model released for public access — available to every plan via the console, API, and CLI
- NewNatural-language infrastructure operations: describe intent and VxThinker plans, previews, and executes with approval gates
- NewContext-aware troubleshooting that reads your logs, metrics, and config to propose root-cause fixes
- NewVxThinker API endpoints with streaming responses and tool calling for embedding into custom workflows
- ImprovedUp to 40% lower latency on multi-step reasoning tasks compared to the private preview build
- ImprovedEnterprise plans gain custom VxThinker fine-tuning on your own runbooks and codebase
- SecurityPer-tenant prompt isolation and zero data retention by default for all VxThinker requests
AI Agents v2 & Kubernetes Federation
Major release with next-gen AI agents capable of multi-step infrastructure operations and cross-cluster Kubernetes federation.
- NewAI Agents v2 with multi-step reasoning, tool chaining, and human-in-the-loop approval workflows
- NewKubernetes multi-cluster federation for global workload distribution and failover
- NewReal-time collaborative code editing with cursor presence and conflict resolution
- ImprovedDashboard load time reduced by 60% with streaming server components
- ImprovedContainer registry now supports OCI artifacts and Helm chart hosting
- FixedFixed WebSocket reconnection drops during extended Cloud Shell sessions
- SecurityUpgraded TLS 1.3 cipher suites across all API endpoints
Migration Engine & Cost Intelligence
- NewAI-powered migration engine with auto-discovery, dependency mapping, and zero-downtime cutover
- NewCost Intelligence dashboard with spend forecasting, anomaly detection, and optimization recommendations
- NewWireGuard-based VPN mesh networking between cloud regions and on-premise sites
- ImprovedTerraform plan preview now shows cost impact estimates before apply
- ImprovedDatabase backups support cross-region replication with configurable retention policies
- FixedResolved intermittent 502 errors during blue-green deployments with large container images
- FixedFixed audit log export failing for organizations with over 100K events
AI Code Completion & Observability Suite
- NewAI-powered code completion in the Cloud IDE with context-aware suggestions from your codebase
- NewUnified observability suite combining metrics, logs, and distributed traces in a single view
- NewNew managed ClickHouse offering for real-time analytics workloads
- ImprovedKubernetes cluster provisioning time reduced from 8 minutes to under 3 minutes
- ImprovedCI/CD pipeline editor with visual DAG builder and parallel stage support
- FixedFixed SSH key rotation not propagating to all nodes in managed Kubernetes clusters
- SecurityAdded FIPS 140-2 validated encryption module for government workloads
Multi-Cloud Governance & SSO
- NewMulti-cloud governance policies with drift detection and auto-remediation across AWS, GCP, and Azure
- NewEnterprise SSO with SAML 2.0, OIDC, and SCIM directory sync for Okta, Azure AD, and Google Workspace
- NewWorkflow automation engine with cron triggers, webhook events, and approval gates
- ImprovedVM provisioning now supports custom cloud-init scripts and user data templates
- ImprovedAPI rate limits increased to 10K requests/min for Enterprise plans
- FixedFixed DNS propagation delays when switching between load balancer configurations
- DeprecatedLegacy v1 API endpoints deprecated — migrate to v2 by June 2026
Cloud Shell & Domain Management
- NewBrowser-based Cloud Shell with persistent storage, tmux support, and pre-installed CLI tools
- NewDomain management with DNS hosting, SSL certificate automation, and DNSSEC support
- NewOracle Cloud and Alibaba Cloud added as managed providers
- ImprovedDark mode redesign with improved contrast ratios meeting WCAG 2.1 AAA standards
- ImprovedObject storage now supports lifecycle policies and intelligent tiering
- FixedFixed GitHub webhook delivery failures during high-traffic deployment windows
- BreakingMinimum Kubernetes version raised to 1.28 — clusters on 1.27 will be auto-upgraded
Looking for older releases? Check the release archive on GitHub.